<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" xmlns:googleplay="http://www.google.com/schemas/play-podcasts/1.0"><channel><title><![CDATA[Blueishprint by Third Vector]]></title><description><![CDATA[Field notes on turning expert know-how into skills an AI can run, with people keeping the calls that matter.]]></description><link>https://blueishprint.thirdvector.io</link><image><url>https://substackcdn.com/image/fetch/$s_!l3hP!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc6c9beb9-d47d-4d1f-be40-e999d56ea1d2_1024x1024.png</url><title>Blueishprint by Third Vector</title><link>https://blueishprint.thirdvector.io</link></image><generator>Substack</generator><lastBuildDate>Sun, 22 Nov 2026 17:51:12 GMT</lastBuildDate><atom:link href="https://blueishprint.thirdvector.io/feed" rel="self" type="application/rss+xml"/><copyright><![CDATA[Third Vector]]></copyright><language><![CDATA[en-gb]]></language><webMaster><![CDATA[blueishprint@substack.com]]></webMaster><itunes:owner><itunes:email><![CDATA[blueishprint@substack.com]]></itunes:email><itunes:name><![CDATA[Third Vector]]></itunes:name></itunes:owner><itunes:author><![CDATA[Third Vector]]></itunes:author><googleplay:owner><![CDATA[blueishprint@substack.com]]></googleplay:owner><googleplay:email><![CDATA[blueishprint@substack.com]]></googleplay:email><googleplay:author><![CDATA[Third Vector]]></googleplay:author><itunes:block><![CDATA[Yes]]></itunes:block><item><title><![CDATA[Founding manifesto: Elevating conversations]]></title><description><![CDATA[The data that matters most for AI in an organisation sits in no system.]]></description><link>https://blueishprint.thirdvector.io/p/founding-manifesto-elevating-conversations</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/founding-manifesto-elevating-conversations</guid><dc:creator><![CDATA[Tim de Rooij]]></dc:creator><pubDate>Thu, 15 Oct 2026 20:22:53 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!LRQz!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!LRQz!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!LRQz!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg 424w, https://substackcdn.com/image/fetch/$s_!LRQz!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg 848w, https://substackcdn.com/image/fetch/$s_!LRQz!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!LRQz!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!LRQz!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg" width="1456" height="1997" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/b0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1997,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:1615094,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/jpeg&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://blueishprint.substack.com/i/220590708?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!LRQz!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg 424w, https://substackcdn.com/image/fetch/$s_!LRQz!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg 848w, https://substackcdn.com/image/fetch/$s_!LRQz!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!LRQz!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb0d57c28-cbc4-44c1-9367-e85f1b8bd27d_2835x3889.jpeg 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><h2>Human conversations</h2><p>When I was in my late teens, there was a time I wanted to be a journalist. The investigative type. The one who can go on a little adventure. Talk to people. Ask questions. Learn something new and connect the dots.</p><p>Journalism was not the path I ended up walking, but what stayed with me is using conversations to learn and understand things better.</p><p>In one of my previous roles, I worked with an executive coach. We explored my intrinsic motivators and &#8220;unique&#8221; capabilities. We assessed feedback from colleagues, and she drilled down to understand my way of working. After a few sessions, she summarised it as &#8220;elevating conversations&#8221;.</p><p>I remember thinking that it sounded uninspiring. Not &#8220;leadership under complex circumstances&#8221;, or &#8220;designing creative solutions for hard problems&#8221;.</p><p>No.</p><p>Elevating conversations.</p><p>Reflecting on it now, I&#8217;ve come to appreciate it more and see how it&#8217;s a helpful capability in this new era of artificial intelligence (AI).</p><p>What&#8217;s more, it sits at the core of my vision for human and AI collaboration, and, by extension, for Third Vector.</p><h2>Machine conversations</h2><p>I started to develop a deep interest in AI in 2014, when I first learned about it during my time at Deloitte Innovation. Coming from a finance background, with few AI people in my network to talk to, I took <a href="https://www.coursera.org/learn/machine-learning">Andrew Ng&#8217;s machine learning course on Coursera</a> in 2016 to better understand the technology. A few years later, when I was working at AI software company <a href="https://www.tamr.com/">Tamr</a>, we deployed supervised <a href="https://en.wikipedia.org/wiki/Machine_learning">machine learning</a> (ML) models to transform and unify massive datasets for large enterprises. With subject-matter experts in the loop training the models, customers could do at scale what had been impossible before. Using this new technology in combination with human expertise, we were unlocking millions in business value.</p><p>With machine learning, humans interacted with the AI model by providing training labels, examples of good and bad answers. The model applied what it had learned to unseen data and, after a couple of training rounds, often performed on par with, or better than, its teachers. The ML model learned from a very basic &#8220;conversation&#8221;.</p><p>Today, with the large language models (LLMs) powering Claude, ChatGPT, Gemini and others, you interact with the AI through language. Besides the training data and model weights, it is the quality of the conversations (or &#8220;prompts&#8221;) that drives the quality of the output.</p><p>We have all had the experience of getting a confident-sounding answer from an AI, only to second-guess it and realise it was flawed after a more thorough reading. Automation bias, the human tendency to over-trust AI-generated answers, is a real thing.</p><p>A high-quality conversation is one where you challenge the AI, stress-test its assumptions and check for logical flaws. If you treat these tools like a question-and-answer box, value is limited. But if you treat them as a collaboration tool, to sharpen your own thinking and capture what you learn, value compounds.</p><h2>Judgement architecture</h2><p>The kind of questioning described in the previous section has been around for more than 2,000 years and is known as the Socratic method.</p><p>From <a href="https://en.wikipedia.org/wiki/Socratic_method">Wikipedia</a>: <em>The Socratic method is a form of argumentative dialogue in which an individual probes a conversation partner on a topic, using questions and clarifications, until the partner is pressed to come to a conclusion on their own, or else their reasoning breaks down and they are forced to admit ignorance.</em></p><p>This type of conversation helps you make better decisions. It improves judgement and makes the reasoning explicit.</p><p>As we hand over more complex and high-stakes tasks to AI, the reasoning behind them needs to be explicit. That explicit version is judgement architecture: how decisions are made, including thresholds, trade-offs and escalation triggers.</p><p>Important decisions are often made by experts who rely on tacit knowledge to apply judgement. Tacit knowledge is what experts know but have never written down, and often can&#8217;t easily put into words. As organisations adopt AI and delegate more to autonomous systems, turning tacit knowledge into explicit judgement architecture becomes a precondition for delegating well.</p><p>Take the example of a Dutch installer of solar panels and heat pumps. Five seasoned experts review every sales order, text and photos, before anything is installed. Does the roof have the right slope for the panels sold? Where should the heat pump go? They either send the order back to the sales rep or turn it into instructions for the installation team. On paper, both are checklist questions. In practice, every roof and every garden is different. Where the heat pump goes usually comes down to distance, because of noise rules, but it also needs enough airflow, and every choice affects how well the system performs. The experts weigh those factors together. A checklist tests them one at a time. Until that judgement is made explicit, the review can&#8217;t be handed to AI without losing what makes it work, and it stays with five people.</p><p>Within ten years, organisations will maintain their judgement architecture the way they maintain their financial records (updated and audited). Only part of that architecture can be found in their systems. System logs, email and chat history, and permission settings show what gets decided and by whom. Conversation is how you find out why.</p><p>In the age of AI, the Socratic method works in new ways. In addition to using it in conversations with humans, you can use it to question the AI, and the AI can use it to question you.</p><h2>Third Vector&#8217;s thesis 1.0</h2><p>Third Vector&#8217;s thesis is this: the data that matters most for AI in an organisation sits in no system. AI can automate large chunks of intelligence, but for complex tasks human experts rely on tacit knowledge: exactly the data no system holds. Drawn out through in-depth conversation, this knowledge becomes explicit judgement architecture, built into skills a whole organisation can run. Experts know which questions to ask. An AI equipped with a skill can ask those questions of everyone who uses it.</p><p>A year ago, the skills I am talking about in the thesis barely existed. Anthropic introduced them as <a href="https://www.anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills">Agent Skills on 16 October 2025</a>. The <a href="https://agentskills.io/home">Agent Skills standard</a> was open-sourced by Anthropic in December 2025 and is now supported by OpenAI&#8217;s ChatGPT and Codex, Google&#8217;s Gemini CLI, GitHub Copilot, Mistral and other AI tools. That makes skills portable: the judgement an organisation captures isn&#8217;t locked to one model or vendor.</p><p>A skill in its most basic form is a text file in .md (<a href="https://en.wikipedia.org/wiki/Markdown">Markdown</a>) format that contains written instructions in plain language. An AI agent reads the file and follows the instructions. A skill is close to a process document. But process documents hardly ever get read, while skills get read every single time an AI runs them.</p><p>Compare that with the supervised models I described earlier. What they learned from training labels ended up in their weights, where nobody could read it. Anyone can read a skill, edit it, argue with it and improve it. That makes authorship possible: expertise becomes a tangible artefact you can own and improve over time.</p><p>Organisations that adopt skills in their way of working create capacity to pursue new, meaningful work. Skills capture specialised knowledge and provide a scalable way to share the expert&#8217;s judgement architecture with the broader team. This enables organisations to move from <a href="https://blueishprint.substack.com/p/from-individual-ai-to-institutional">individual AI to institutional AI</a>: the difference between each person&#8217;s private chats and shared skills the organisation owns and improves.</p><h2>Intense curiosity</h2><p>AI-native organisations need explicit judgement architectures. Where judgement is tacit, conversation is how you capture it. Skills are how the organisation runs it, with the expert keeping authorship and gaining reach.</p><p>Third Vector&#8217;s name holds the idea. A vector carries two things: magnitude and direction. And we are in the third wave of AI. First it was predictive (machine learning), then generative (LLMs), and now agentic, where AI systems make decisions on their own. That gives enormous magnitude: speed, scale, autonomy. But magnitude without direction just gets you to the wrong place faster. Inside the models, ideas are stored as vectors, and similar ideas point in the same direction. In organisations, direction comes from human judgement.</p><p>I&#8217;m building Third Vector with intense curiosity about applying AI, understanding and scaling expert know-how, and creating methods to capture tacit judgement. There is so much to discover. The investigative journalist in me gets his little adventure after all.</p><p>The approach is <em>elevating conversations</em>: in interviews with experts, and in skills designed to apply expert judgement and stimulate critical thinking when people engage with the AI.</p><p>I&#8217;m looking forward to our first conversation.</p><div><hr></div><p><em>Photo by <a href="https://www.pexels.com/@david-underland/">David Underland</a> on Pexels. Thank you, David.</em></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en-gb&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Blueishprint by Third Vector! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div>]]></content:encoded></item><item><title><![CDATA[From Individual AI to Institutional AI]]></title><description><![CDATA[Non-technical domain experts are building real software with Claude. The next step is harder, and more important.]]></description><link>https://blueishprint.thirdvector.io/p/from-individual-ai-to-institutional</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/from-individual-ai-to-institutional</guid><dc:creator><![CDATA[Tim de Rooij]]></dc:creator><pubDate>Thu, 28 May 2026 06:53:27 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!ZB6m!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!ZB6m!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!ZB6m!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg 424w, https://substackcdn.com/image/fetch/$s_!ZB6m!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg 848w, https://substackcdn.com/image/fetch/$s_!ZB6m!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!ZB6m!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!ZB6m!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg" width="1456" height="971" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:971,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:4562659,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/jpeg&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://thirdv3ctor.substack.com/i/199448916?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!ZB6m!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg 424w, https://substackcdn.com/image/fetch/$s_!ZB6m!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg 848w, https://substackcdn.com/image/fetch/$s_!ZB6m!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!ZB6m!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F69f1fdda-e82a-4558-8ccf-ee97d74388e5_5760x3840.jpeg 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p>Remco runs payroll at a B2B company with a remote workforce of more than 1,500 people across dozens of countries. Different rules in every jurisdiction. Multiple currencies. People flowing in and out of roles every week. And one immovable constraint: when payroll breaks, people can&#8217;t pay their rent.</p><p>Every month, Remco reconciles the numbers from the local employers of record with what&#8217;s sitting in SAP. Every month, he reformats datasheets to match the system&#8217;s expectations. Until recently, that work cost him and his team several days a month of careful, error-prone manual processing.</p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p>Now it doesn&#8217;t. Remco has built a small set of Claude apps running locally on his machine that do most of it for him. He&#8217;s not a developer. He has never written production code. But he has <em>deep domain knowledge</em> about how his payroll process actually works, and he used that knowledge, plus a long Claude chat, to vibe-code his way to something that runs. The lack of internal dev resources is no longer the constraint it used to be.</p><p>That&#8217;s the part of this story everyone is talking about. It&#8217;s also where most people stop.</p><h2>The pattern: powerful, but private</h2><p>Remco is not the only one. In nearly every organization I work with, I see a version of him: someone in finance, or ops, or HR, or support, who has quietly built themselves an app that automates a real, gnarly workflow. The app works. The person is more productive. From the outside, it looks like the AI dream realized.</p><p>Look closer, though, and you see the problem. These apps live in a single user&#8217;s account. They sit behind one person&#8217;s login, in one person&#8217;s chat history, with no version control, no documentation, no second pair of eyes on the prompts and logic that drive them. The institutional knowledge is in someone&#8217;s head. The app that encodes that knowledge is on someone&#8217;s laptop.</p><p>I call this <em>individual AI</em>. It&#8217;s a real productivity gain, and it&#8217;s also a fragile one.</p><p>If Remco gets sick, the team falls back to days of manual reconciliation. If Remco leaves, the app effectively leaves with him. If his prompt logic has a subtle bug, no one will catch it. If audit asks how a number was produced, &#8220;Claude did it&#8221; is not a satisfying answer. And if three other people on the team have similar needs, they each rebuild the same thing from scratch.</p><p>This is the same story enterprises lived through with spreadsheets in the nineties and with Power BI dashboards in the twenty-tens. Powerful local tools, indispensable to the people who built them, invisible to everyone else. Eventually the organization gets serious and brings them under proper governance. AI is on the same trajectory, except the cycle is moving much faster.</p><h2>The real question</h2><p>So the interesting question is not &#8220;can a domain expert build a useful AI app?&#8221; That one is settled. The interesting question is:</p><p><em>How do you turn an individual AI tool into an institutional one, something the whole team can use and IT can help maintain, without killing the thing that made it valuable in the first place?</em></p><p>This is the conversation I had with Remco. Here is the path we worked out.</p><h2>A four-step path from individual to institutional</h2><p>One precondition before any of this: make sure the process you&#8217;re about to scale is worth scaling. Automate a flawed workflow and you&#8217;ve made it faster, harder to change, and more expensive to undo. With Remco, the reconciliation logic had been refined over years of manual iteration. That history effectively did the stress-testing for him. If your process doesn&#8217;t have that kind of history, do the work first. Walk Claude through your current flow step by step, tell it you&#8217;re about to move from manual to AI-native, and ask it to challenge each assumption and propose improvements. Then come back to step 1.</p><h3>Step 1 (optional but recommended): redesign the user experience</h3><p>The v1 of a vibe-coded app is usually shaped around its builder. The labels make sense to them, the flow assumes their context, the edge cases are the ones they happened to hit. That&#8217;s fine when you&#8217;re the only user. The moment you have more, it isn&#8217;t.</p><p>Open the app in claude.ai/design (currently browser-only, not yet available in the Claude desktop app), give it the brief that the <em>functional logic must stay intact</em>, and ask it to optimize for someone who didn&#8217;t build it. Then export the Claude Code hand-off document, a detailed development briefing that captures the intent, the data flows, and the design decisions in a form a developer (or a coding agent) can actually work from.</p><p>This step is optional, but skipping it is usually a false economy. UX is what turns &#8220;Remco&#8217;s tool&#8221; into &#8220;the payroll team&#8217;s tool.&#8221;</p><h3>Step 2: get a real repository</h3><p>Request access to your organization&#8217;s GitHub (or GitLab, or Azure DevOps, whatever your IT team already runs) and create a repository for the app. Nothing about this step is glamorous. It&#8217;s also the step that unlocks everything else: version history, code review, branch protections, secrets management, the ability for someone other than you to contribute.</p><p>If your IT team asks why an HR or payroll person needs a repo, the honest answer is: <em>because the thing I built is now software, and software belongs in a repo.</em> That framing tends to go over well.</p><h3>Step 3: clone the repo locally</h3><p>Pull the empty repo down to your machine. Claude Code can walk you through this end-to-end if it&#8217;s your first time: installing git, authenticating, setting up the local folder, the works. Treat this as the moment your project graduates from a chat tab to a real workspace on disk.</p><h3>Step 4: rebuild v2 in Claude Code</h3><p>Start a new Claude Code session pointed at the local repo. Hand it three things: the hand-off document from step 1, the current code artefacts from your v1 app, and a clear statement of intent for v2. Then iterate.</p><p>Claude Code is now writing into a versioned, reviewable codebase rather than a transient chat. Every change is tracked. Every decision is recoverable. And, the part people underestimate, you now have something an IT colleague can plug into when you need help with the genuinely hard parts: authentication, database connections, secrets, deployment, scheduled jobs.</p><h2>What you actually get</h2><p>Promoting an app from individual to institutional sounds bureaucratic, but each piece earns its keep:</p><ul><li><p><strong>Version control.</strong> Git gives you a free time machine, a built-in review process, and a way to roll back the change that broke production at 17:55 on a Friday. Vibe-coded apps that live in a chat have none of this.</p></li></ul><ul><li><p><strong>A shared place for documentation.</strong> README files, architecture notes, prompt explanations, runbooks. The institutional knowledge that used to live in Remco&#8217;s head gets written down where the next person can find it.</p></li></ul><ul><li><p><strong>A clean handoff surface with IT.</strong> Most non-developers can get an app 70% of the way there with Claude. The last 30%, the parts involving real auth, real data, real production concerns, is where you genuinely want someone with deeper engineering experience. A repo is the artifact your IT team knows how to engage with. Without it, every conversation starts from &#8220;show me what you&#8217;ve got&#8221; and ends in frustration.</p></li></ul><ul><li><p><strong>A path to wider adoption.</strong> Once the app is in a repo, deploying it for the whole team is a known problem. Internal hosting, SSO, role-based access: IT does this every day. Asking them to &#8220;deploy your Claude chat for the team&#8221; is a non-starter. Asking them to deploy a small internal app from a repo is just Tuesday.</p></li></ul><ul><li><p><strong>Defensibility.</strong> When audit, security, or legal eventually come knocking, and they will, you have something to show them. Code. History. Reviewers. Documentation. The same defensibility the company expects from any other piece of software it depends on.</p></li></ul><h2>Not everything needs to graduate</h2><p>A quick caveat, because I don&#8217;t want to oversell this. Not every Claude app deserves the trip to GitHub. Some are genuinely one-off: a personal research helper, a one-time data cleanup, a draft generator that only you will ever use. Promoting those is overhead with no upside.</p><p>A useful filter: an app is a candidate for institutional promotion when at least two of the following are true. It runs on a recurring cadence (monthly close, weekly report). More than one person could benefit from it. It touches data or decisions that the company would want an audit trail for. Losing the person who built it would be operationally painful.</p><p>Remco&#8217;s payroll reconciliation hits all four. That&#8217;s why we&#8217;re moving it.</p><h2>Why this matters now</h2><p>The reason this conversation is happening at all is that the bottleneck has moved. Two years ago, the constraint on internal automation was <em>can we build it?</em> and the answer was usually &#8220;not without a dev team we don&#8217;t have.&#8221; Today, the constraint is <em>can we operate it at the standard the organization needs?</em> That&#8217;s a different problem, and it&#8217;s not one Claude alone solves for you. It&#8217;s a problem of process, ownership, and the boring infrastructure of software in production.</p><p>The companies that figure this out first will have a quiet but compounding advantage. Not because they have better AI, everyone has access to the same models, but because they have more <em>institutionalized</em> AI. Apps that survive the person who built them. Workflows that improve every month instead of degrading the moment someone gets sick. A pipeline that takes the best ideas from the Remco&#8217;s of the organization and turns them into shared infrastructure.</p><p>Individual AI gets you the demo. Institutional AI gets you the operating leverage.</p><p>If you have a Remco on your team, the next move isn&#8217;t to celebrate what they built. It&#8217;s to give them the path to make it bigger than themselves.</p><p></p><p><em>Image credit: Markus Spiske (via Pexels)</em></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div>]]></content:encoded></item><item><title><![CDATA[How to set up an outbound motion for under €300 a month]]></title><description><![CDATA[A solo founder's automated outreach blueprint, end to end.]]></description><link>https://blueishprint.thirdvector.io/p/how-to-set-up-an-outbound-motion</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/how-to-set-up-an-outbound-motion</guid><dc:creator><![CDATA[Friso Paping]]></dc:creator><pubDate>Fri, 22 May 2026 13:00:19 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!FRJL!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!FRJL!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!FRJL!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg 424w, https://substackcdn.com/image/fetch/$s_!FRJL!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg 848w, https://substackcdn.com/image/fetch/$s_!FRJL!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!FRJL!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!FRJL!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg" width="1456" height="1941" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/b6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1941,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:543207,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/jpeg&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://thirdv3ctor.substack.com/i/198835788?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!FRJL!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg 424w, https://substackcdn.com/image/fetch/$s_!FRJL!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg 848w, https://substackcdn.com/image/fetch/$s_!FRJL!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!FRJL!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6875854-5e89-47a2-b66e-260d450205d5_3848x5131.jpeg 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p></p><p>The headcount math for outbound used to look something like this. Three SDRs writing sequences, two EAs cleaning lists and routing replies. Five people, fifty thousand a month all-in once you add tools. A real team for a real budget.</p><p>I run the same motion as one person for &#8364;300 a month.</p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p>This isn&#8217;t a productivity hack. It&#8217;s a stack. Six tools, one operator, and a clear separation of what humans do versus what software does. Below is the full blueprint. The process, the systems, the costs, where it breaks.</p><p>If you&#8217;re a founder who needs to do outbound but can&#8217;t justify a team yet, this is the playbook I use.</p><h2>The stack</h2><p>The full stack costs around &#8364;60 a month for the bare minimum and &#8364;300 if you add enrichment.</p><ul><li><p><strong>Instantly.ai</strong> for sequencing, sending, and reply routing. &#8364;37/month.</p></li><li><p><strong>A cold-sending domain</strong> separate from your main domain, so deliverability stays clean. &#8364;15/year, call it &#8364;1.25/month.</p></li><li><p><strong>Three Google Workspace mailboxes</strong> on the cold domain to spread sending volume. &#8364;18/month total.</p></li><li><p><strong>Claude (Pro or API)</strong> for drafting variants, scoring segments, and configuring campaigns. &#8364;20-200/month depending on usage.</p></li><li><p><strong>Cal.com or similar</strong> for the meeting link. Free tier works.</p></li><li><p><strong>Clay (optional)</strong> for enrichment if you want fancier targeting. &#8364;149/month.</p></li></ul><p>The core minimum without Clay lands around &#8364;60-100 a month for technology. The rest is your time.</p><h2>The blueishprint</h2><p>This is a seven-step B2B outbound email campaign motion:</p><ol><li><p><strong>ICP definition</strong> &#8212; who exactly you&#8217;re targeting.</p></li><li><p><strong>General pain mapping</strong> &#8212; what hurts enough to make them open a cold email.</p></li><li><p><strong>Segment scoring</strong> &#8212; which slices of the ICP are worth a lane.</p></li><li><p><strong>Lane and lead selection</strong> &#8212; the two or three segments you&#8217;ll actually run, and the leads in each.</p></li><li><p><strong>Message variants</strong> &#8212; the angles you&#8217;ll test per lane.</p></li><li><p><strong>Segment-specific personalization</strong> &#8212; how each variant lands for its specific audience.</p></li><li><p><strong>Set up and run the campaign</strong> &#8212; the operational engine that turns all of the above into sent email and booked meetings.</p></li></ol><p></p><p>Steps 1 through 6 are strategy and content design. The output is a plan: a scored segment list, two chosen lanes, message variants ready to ship.</p><p>Step 7 is execution, and it&#8217;s where this post lives. The next section provides the detailed walkthrough, from locking in your lanes (which is where the strategic prep crystallizes into an actual campaign), through setting up sending infrastructure, configuring cadences, building lists, sending, and routing replies. After the campaign starts, every reply routes into one of six buckets that determine the follow-up path.</p><p>Three recommendations to make this work as a single-operator motion.</p><ul><li><p>Commit to lanes. Two or three segments at a time, never more. Each lane gets its own campaign, leadlist, and message variants.</p></li></ul><ul><li><p>Run experiments inside experiments. The lane tests whether the segment is right. The variants inside the lane test which message resonates. One variant tells you whether the lane works. Three variants tell you why.</p></li></ul><ul><li><p>Let Claude do the assembly work. Drafting, scoring, formatting, configuring. You stay in the loop on judgment calls &#8212; does this segment have the pain we think it has, does this list look real, is this subject line saying anything new. The typing is automated.</p></li></ul><p>Sub-steps 2 through 4 of the walkthrough run in parallel during the two-week domain warmup, so calendar time is about three weeks before your first real email goes out.</p><h2>The campaign setup</h2><h3>Step 1: Score segments and pick two lanes</h3><p>This is the only purely strategic step. Skip it and you&#8217;ll burn the rest of the stack on the wrong audience.</p><p>I keep a scoring sheet with around 45 segments scored across ten weighted dimensions: pain, PII handling, volume, reachability, ability to pay, ACV potential, K-factor, regulatory pressure, segment homogeneity, and competitive density. Weights are tuneable, formulas update live. The output is a top-10 with a tie-breaker.</p><p>Pick two or three lanes from the top, never more. Three is the upper bound for a solo operator. Above three, you lose the thread on what&#8217;s working.</p><p>For each lane, write the ICP in one sentence. Something like &#8220;independent professional services firm, one to five operators, regulated by the relevant authority, member of the trade body.&#8221; That sentence becomes your filter for everything downstream.</p><h3>Step 2: Set up the cold domain (start warming immediately)</h3><p>This is the part most people get wrong, and it&#8217;s the slowest piece, so start it on day one.</p><p>Sending cold email from your main domain risks torching its reputation. Bounces, spam complaints, and unusual sending patterns will tank deliverability across all your real correspondence too. Your investor updates start landing in spam folders.</p><p>The fix is a separate cold-sending domain. Volume only ever leaves the secondary, the main domain stays reserved for real correspondence.</p><p>On the cold domain you set up:</p><ol><li><p>DKIM, SPF, DMARC authentication. Without these your mail goes straight to spam. Instantly walks you through this in setup.</p></li><li><p>Three mailboxes minimum on Google Workspace. Sending volume should never exceed about 30 emails per mailbox per day in warmup, scaling to 50 once warm. Three mailboxes give you 90-150/day capacity.</p></li><li><p>Two weeks of warmup before any real sending. Instantly has a built-in warmup tool that simulates real conversation across a network of mailboxes. Skip this and your first real campaign will land in spam universally.</p></li></ol><p>While the warmup runs in the background, do Steps 3 and 4 in parallel.</p><h3>Step 3: Configure cadences (let Claude drive)</h3><p>This is the step where the time math actually changes.</p><p>Setting up an Instantly campaign by hand takes about two hours of clicks. Four sequences, four steps each, sixteen email bodies, sixteen subject lines, plus delays and sender mailboxes. For a real campaign with proper variants, multiply that by three: six hours of clicking, copying, pasting, double-checking.</p><p>I let Claude drive the browser. End to end the same campaign takes about an hour, most of which is me writing the cadence in a spreadsheet and watching Claude work.</p><p><strong>The workflow</strong></p><p>I write the cadence in a single sheet, one row per email step. Columns:</p><ul><li><p>sequence_name &#8212; e.g. &#8220;Lane A &#8212; Problem-first&#8221;</p></li><li><p>step &#8212; 1 through 4</p></li><li><p>subject &#8212; the subject line</p></li><li><p>body &#8212; the email body, with {{first_name}} and other variables inline</p></li><li><p>wait_days &#8212; days to wait before this step fires (0 for the first email)</p></li></ul><p>For a three-variant lane that&#8217;s 12 rows. For two lanes running three variants each, 24 rows. The sheet is the source of truth. If you want to change a subject, you change the cell, not the campaign.</p><p>Then I open Instantly in Chrome, share the sheet with Claude, and give it a one-paragraph prompt: &#8220;Create a new campaign called X, add three sequences named per the sequence_name column, and for each sequence add the four steps from the sheet with their subjects, bodies, and wait days.&#8221; Claude navigates Instantly&#8217;s UI, creates the campaign, and configures every step from the sheet.</p><p><strong>The {{ trick</strong></p><p>The non-obvious part is how Claude actually gets the email body into the editor.</p><p>Instantly&#8217;s body editor &#8212; like most rich-text editors built for outreach tools &#8212; intercepts {{ as a variable trigger and pops a dropdown. If Claude tries to type the body in naturally, every variable opens the picker and the text gets mangled. Paste doesn&#8217;t work either; the editor reformats on paste.</p><p>The workaround: switch the editor to HTML / code view. That view is a plain textarea, no interception. Claude uses its JavaScript tool to set the textarea&#8217;s value directly, then flips back to rich-text view. Instantly re-parses the body, variables render as proper merge fields, and the test send works exactly as expected.</p><p>This pattern works in most of the outbound and CRM tools I&#8217;ve tried. Wherever the rich editor blocks you, the code view is the escape hatch.</p><p><strong>Where the human stays</strong></p><p>I watch Claude work, not the whole hour, but enough to catch the failure modes. Sometimes Claude clicks the wrong &#8220;create&#8221; button if there are two on the page. If a sequence_name in the sheet has a typo, Claude creates a typo&#8217;d sequence. And test sends are mandatory before you go live: five minutes per sequence catches variable mismatches, escaped HTML, and the occasional case where Claude pasted the wrong cell.</p><p>Per lane I run three message types in parallel: a problem-first variant, a founder-voice variant, and a price-or-time variant. Each runs as its own four-step sequence with the same day cadence, day 0, day 3, day 7, day 12. The variants compete head-to-head inside the lane. By the end of week two you know which voice is landing.</p><h3>Step 4: Build the lead list</h3><p>Lead sourcing can go fully automated (Clay, Apollo, Lusha, and similar) or fully manual. I recommend starting manual. You&#8217;ll learn what a good lead actually looks like in your segment, and that knowledge is what makes the automation work later. A follow-up post will cover the manual sourcing playbook in detail.</p><p>Either way, this is the step where you have to think.</p><p>Tools won&#8217;t do this for you, and shouldn&#8217;t. The lead list is your judgment about who you actually want to reach. If you outsource it to a generic enrichment platform or a cold AI agent, you&#8217;ll get a list that looks plausible and contains nobody who matters.</p><p>The pattern I use:</p><ol><li><p>Find a primary list from public registers and trade association directories. LinkedIn Sales Navigator works if you have it. The cheapest leads are usually the cleanest.</p></li><li><p>Manually scrub. Open the list, read the names, drop anyone obviously wrong. This takes an hour for 100 leads. It&#8217;s the most valuable hour in the whole process.</p></li><li><p>Enrich what&#8217;s missing. First name, company name, role title, email. If your source already has it all, skip enrichment.</p></li><li><p>Upload as CSV to Instantly. Map columns to email, first_name, company_name. Set custom fields if you want segment-specific personalization later.</p></li></ol><p>Don&#8217;t skip the manual scrub. Any list of more than 200 leads gets sampled before it ships. If 5% of the sample looks wrong, the list goes back for review.</p><h3>Step 5: Send slow, measure, iterate</h3><p>When you launch, send slowly. Twenty leads a day, not two hundred. You&#8217;re not optimizing for volume yet. You&#8217;re optimizing for whether the message lands at all.</p><p>Two lanes I ran recently make the case. The first lane: around 60 sends, roughly 7% reply rate. A clean signal that the lane works. The second lane: around 80 sends, just over 1%. Same stack, same operator, same week. The lane just didn&#8217;t connect.</p><p>That gap is exactly why you run two or three lanes in parallel. Until you have real reply data, you don&#8217;t know which segment your message will land in. If you&#8217;d bet everything on the second lane, you&#8217;d have spent weeks chasing a signal that wasn&#8217;t there.</p><p>Within each lane I A/B test subject lines and bodies. The winners are rarely the ones I expected. In one sequence, the breakup email &#8212; subject &#8220;Are you OK?&#8221; &#8212; pulled more replies than two carefully polished subjects earlier in the sequence combined. Some subject lines just punch above their weight. Body copy is the same story: the price-first variant beat the founder-voice variant in one lane, then lost in another. The only way to find the winner is to ship three variants and let the data choose.</p><p>Three numbers matter:</p><ul><li><p><strong>Reply rate.</strong> Below 1%, something is wrong with the list, the message, or deliverability. 5 to 15% is good for a tight lane.</p></li><li><p><strong>Meeting rate.</strong> Of replies, how many turn into a 20-30 minute call. 30 to 40% is normal.</p></li><li><p><strong>Bounce rate.</strong> Below 2% is healthy. Above 5% means your list is dirty.</p></li></ul><p>Don&#8217;t track open rate as a primary metric. Apple Mail and Gmail iOS preload images, marking emails as opened whether the recipient actually looked or not. It&#8217;s noise.</p><p>After a week of data, kill the worst variant. Double the volume on the winner. After two weeks, if the numbers hold, expand the leadlist. After a month, you have enough signal to either scale or pivot the lane.</p><p>Most people skip the patience and dump 500 leads on day one. They get bad data, can&#8217;t tell which variant won, and conclude the channel doesn&#8217;t work. The channel works. Patience is the part most people miss.</p><h3>Step 6: Route the replies</h3><p>This is what separates outbound that scales from outbound that vacuums up your time. Every reply lands in one of six buckets, each with its own follow-up path.</p><ul><li><p><strong>Interested</strong> &#8212; connect on LinkedIn, ask for the call directly, follow up after 1, 3, and 7 days if no scheduling response. After three follow-ups without a calendar invite, mark as dead.</p></li><li><p><strong>More info</strong> &#8212; connect on LinkedIn, send content (case study, one-pager) on the same 1/3/7-day cadence.</p></li><li><p><strong>Wrong person</strong> &#8212; custom response asking for the right contact. Same 1/3/7-day follow-up if no name comes back.</p></li><li><p><strong>Contact later</strong> &#8212; custom response acknowledging the timing, then drop the lead back into the sequence after a 3-6 month cooldown.</p></li><li><p><strong>Not interested / auto-reply / do-not-contact</strong> &#8212; dead immediately. Don&#8217;t argue.</p></li><li><p><strong>No response after the full 4-email sequence</strong> &#8212; one final 1-week follow-up, then a 2-week follow-up, then dead.</p></li></ul><p>Tag every incoming email with the right inbox label as you read it. Instantly automates the rest from there: the follow-up sequences, the LinkedIn handoff, the cooldown timer for &#8220;contact later&#8221; leads. Without routing, every reply becomes a 5-minute decision. With routing, it&#8217;s 30 seconds. At scale that&#8217;s the difference between half an hour a day on outbound and three hours.</p><p></p><h2>Where this breaks</h2><p>Three honest limits.</p><p>It doesn&#8217;t work if your ICP is too narrow. If you&#8217;re targeting fifty companies in the world, automated outreach is the wrong tool. Hand-write those fifty.</p><p>It doesn&#8217;t work without a clear hook per segment. Generic value props get generic responses, which is to say, none. The lane-and-variant structure exists because each segment needs its own pain and language.</p><p>It doesn&#8217;t replace a real SDR for complex multi-stakeholder enterprise sales. For a six-figure ACV deal with five stakeholders across procurement, security, and the buying team, you still want a human running the play. This stack is for the segment of business where one cold email leads to one demo and one deal in three weeks.</p><p></p><h2>What it costs in time</h2><p>Around 2 days of active work the first time you set this up end to end. Domain config, lead sourcing, cadence build, mailbox warmup setup, the manual scrub on your first lead list. The 2-week domain warmup runs in the background, so calendar-time it&#8217;s about three weeks before you send your first email. After your first campaign is live, count on 6-8 hours per new campaign. The heavy first-time setup amortizes across every lane after.</p><p>Once it&#8217;s running, ongoing time is maybe 30 minutes a day. Read the replies, route them with inbox labels, drop the polite-nos, mark wrong-people for cleanup. The whole motion fits between morning coffee and the first standup.</p><p>For &#8364;60 to &#8364;300 a month plus a part-time hour a day, you&#8217;re running an outbound function that produces real meetings. That&#8217;s what makes it work as a solo founder motion.</p><p>If your business has a repeatable wedge that works for a defined segment, this stack will produce results that justify the cost within sixty days. If your wedge isn&#8217;t clear yet, this stack will surface that for you fast, and that&#8217;s worth the cost too.</p><div><hr></div><h2>References</h2><ul><li><p><a href="https://www.figma.com/board/RU26KHNjSD053ruxtTJZVm/Outbound-Engine?node-id=0-1&amp;t=5H1h1g5qq6vMneBr-1">Outbound Engine board on Figma</a> &#8212; full process map including reply-routing</p></li><li><p><a href="https://instantly.ai">Instantly.ai</a></p></li><li><p><a href="https://www.anthropic.com/news/claude-for-chrome">Claude in Chrome</a></p></li><li><p><a href="https://cal.com">Cal.com</a></p></li><li><p><a href="https://clay.com">Clay</a></p></li></ul><div><hr></div><p><em>Originally posted to Third Vector&#8217;s The Blueishprint. If you build outbound and want to compare notes, hit reply.<br><br>Image credits: Kindel Media via Pexels</em></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div>]]></content:encoded></item><item><title><![CDATA[We Changed Our Pricing 4 Times in 6 Months. Here's What We Learned.]]></title><description><![CDATA[We built a product, launched it, and immediately got pricing wrong. Then we got it wrong again. And again. Here's the framework we wish we'd started with.]]></description><link>https://blueishprint.thirdvector.io/p/we-changed-our-pricing-4-times-in</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/we-changed-our-pricing-4-times-in</guid><dc:creator><![CDATA[Friso Paping]]></dc:creator><pubDate>Mon, 04 May 2026 17:45:37 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!3m0M!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!3m0M!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!3m0M!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg 424w, https://substackcdn.com/image/fetch/$s_!3m0M!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg 848w, https://substackcdn.com/image/fetch/$s_!3m0M!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!3m0M!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!3m0M!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg" width="1456" height="971" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:971,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:1772496,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/jpeg&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://thirdv3ctor.substack.com/i/196449998?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!3m0M!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg 424w, https://substackcdn.com/image/fetch/$s_!3m0M!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg 848w, https://substackcdn.com/image/fetch/$s_!3m0M!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!3m0M!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F26616484-e221-420b-9d16-405c448ce69d_6512x4341.jpeg 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-GK1x3M" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-GK1x3M" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p>Three weeks after launch, I sat in a meeting with a potential partner and watched their face go blank when I explained our pricing. Not confused. Blank. The kind of blank that means &#8220;I have no idea how this maps to my business, and I&#8217;m too polite to say so.&#8221;</p><p>We&#8217;d spent months building tiered pricing that made perfect sense from our side. Four tiers, clean feature gates, volume discounts. It was elegant. It was logical. And it was completely disconnected from how our customers actually think about value.</p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p>That meeting was the first of four pricing rewrites in six months. Each one taught us something we couldn&#8217;t have learned from a spreadsheet alone.</p><h2>The mistake everyone makes (including us)</h2><p>There are really only three ways to justify a price: <em>cost-based</em> (mark up what it costs you to deliver), <em>competition-based</em> (anchor to what the market already pays), or <em>value-based</em> (charge a fraction of the value or savings you create). Most founders default to the first one. We did too.</p><p>The problem with cost-plus pricing is that your customer doesn&#8217;t care what it costs you to deliver. They care what it&#8217;s worth to them. And the gap between those two numbers is where all the interesting pricing decisions live.</p><p>When we first priced the product, we started with our infrastructure cost per identity check, added a margin, and arrived at a per-check price. Technically correct. Strategically useless. Because a per-check price means nothing to a wealth management firm trying to figure out whether your platform will save them money on client onboarding.</p><h2>Version 1: Per-check pricing (the logical mistake)</h2><p>Our first model was simple: charge per identity verification. Max 10 cents per check, with volume discounts stepping down to 8 cents above 100,000 monthly checks. Four tiers &#8212; Free, Standard, Premium, Enterprise &#8212; gated by features and volume.</p><p>On paper, this was clean. In practice, nobody could calculate what it would actually cost them. A wealth manager with 80 active client connections running periodic data refreshes would need to estimate their monthly check volume, factor in the tier they&#8217;d land on, account for overages, and somehow arrive at a budget number. Too much mental math, too much uncertainty.</p><p>The feedback wasn&#8217;t &#8220;your price is too high.&#8221; It was silence. People couldn&#8217;t engage with the model at all.</p><h2>Version 2: Per-user pricing (the overcorrection)</h2><p>So we swung the other way. Flat monthly fee per user. Simple to understand, easy to budget for. But it created a different problem: it penalized our best customers. A firm with 350 wealth managers would pay 350x regardless of how much value they were extracting. Heavy users and light users paid the same. The firms that would benefit most were the ones getting the worst deal.</p><p>We also couldn&#8217;t differentiate between a small advisory practice with 25 wealth managers and a large institution with 350. The value per user was completely different across these segments, but our pricing pretended it was the same.</p><h2>Version 3: Value-based, but we calculated the value wrong</h2><p>This is where things got interesting. We decided to flip the model: instead of pricing based on what it costs us, price based on what it saves them. Calculate the actual cost of onboarding and maintaining a client today (hourly wage &#215; time per event &#215; external costs), calculate the same cost with our product, and the difference is the value we create.</p><p>The framework was right. The execution was wrong. We were calculating total value across the entire firm and then trying to capture a percentage. But the buyer wasn&#8217;t thinking in &#8220;total firm savings.&#8221; They were thinking &#8220;what does this cost me per client?&#8221;</p><p>We&#8217;d built a top-down model when the buyer needed a bottom-up story.</p><h2>Version 4: Bottom-up value pricing (what actually works)</h2><p>The breakthrough came when we rebuilt the model from the smallest unit up:</p><p><strong>Step 1: Value per single client.</strong> What does it cost to onboard and maintain one client today? What does it cost with our product? The difference is the value per client. For a typical wealth manager, current onboarding cost was roughly &#8364;45 per client (blended hourly wage &#215; time &#215; external verification costs). With our product, that dropped to around &#8364;12. Value created: &#8364;33 per client.</p><p><strong>Step 2: Value per user.</strong> Multiply by average clients per user. A wealth manager at a small firm handles about 50 clients. At a large institution, closer to 200. So value per user ranges from &#8364;1,650 to &#8364;6,600 annually, depending on segment.</p><p><strong>Step 3: Value capture percentage.</strong> We don&#8217;t capture all the value. We capture a share that feels fair relative to what the customer keeps. We landed on different capture rates by segment: higher for enterprise (they get more value per user), lower for small firms (keep the barrier low).</p><p><strong>Step 4: Sense check.</strong> Does the resulting monthly fee per user feel reasonable? Can a buyer look at it and immediately understand the ROI without a spreadsheet? If the answer to both is yes, the pricing works.</p><p>This model let us walk into a meeting and say: &#8220;Your onboarding cost per client drops from &#8364;45 to &#8364;12. You have 100 wealth managers averaging 100 clients each. Here&#8217;s exactly what that saves you, here&#8217;s what we cost, and here&#8217;s your net ROI.&#8221; One slide. No confusion. No blank faces.</p><h2>Triangulate: cost is the floor, competition is the anchor, value is the ceiling</h2><p>Looking back, the deeper lesson isn&#8217;t that value-based pricing is the right answer. It&#8217;s that no single method gives you a defensible price on its own.</p><p>We now triangulate three lenses every time we revisit pricing:</p><p><strong>Cost-based answers: what&#8217;s the floor?</strong> Sum what it costs to serve one customer at scale (infrastructure, support, amortized dev, payback on acquisition cost). Add the margin you need. That&#8217;s your sanity check, not your target. If your price is below this number, you&#8217;re choosing to subsidize, and you should know exactly why.</p><p><strong>Competitive answers: what does the market already accept?</strong> Benchmark adjacent products. This anchors your pricing to what customers already pay elsewhere, which is where their reference point lives whether you like it or not. The risk is picking the wrong peer set; pick three peer groups, not one.</p><p><strong>Value-based answers: what&#8217;s the most we could charge?</strong> Quantify time saved, risk avoided, costs eliminated. Apply a willingness-to-pay rate (10&#8211;15% is a common starting point for new products). This gives you the ceiling, the most defensible long-term story for sales, and the ROI math the buyer will use internally.</p><p>The right price is somewhere in the overlap. When we first priced the product, we used only one lens (cost) and ended up with a number that was technically defensible and commercially inert.</p><h2>Pick the right structure (and don&#8217;t be afraid of a hybrid)</h2><p>Once you have a price, you have to decide how to charge for it. Per seat? Per unit of usage? Flat platform fee?</p><p>The honest answer is that almost everyone interesting ends up with a hybrid. We&#8217;ve recently been working with an AI platform serving customer support teams that landed on a <em>two-part tariff</em>: a flat monthly platform fee that gets you access and unlimited users, then per-unit consumption rates for the modules you actually use (per word translated, per QA review). Volume tiers bring the per-unit rate down as customers scale. A 10% multi-module discount nudges expansion. Annual commitments get a further discount in exchange for predictability.</p><p>It works because each piece does a different job. The platform fee establishes a stable relationship and covers fixed costs. The consumption fee aligns what the customer pays with what they get. The volume tiers reward growth. The annual discount pulls cash forward and reduces churn risk.</p><p>If you want to push this further, three-part tariffs (platform fee + bundled usage + per-unit overage) tend to capture even more value. Behavioral research suggests customers on a three-part tariff increase their usage by around 15% versus less than 1% for those on a pure two-part tariff. The bundled allowance changes how they think about consumption.</p><p>This matters more than ever for AI products. When the buyer&#8217;s mental model is &#8220;I&#8217;m replacing seats with agents,&#8221; your pricing structure has to make that math obvious. A pure per-seat model under-prices the actual leverage. A pure consumption model gives the buyer pricing anxiety. Most teams we see end up somewhere in between, with a platform fee that establishes the relationship and consumption-based pricing that aligns with the value delivered per ticket, per interaction, or per outcome.</p><h2>What we actually learned</h2><p><strong>Start from the customer&#8217;s smallest unit of value.</strong> Don&#8217;t model total firm savings. Model the saving per client, per transaction, per ticket, per whatever your customer naturally thinks in. Then build up. People can intuitively validate small numbers (&#8221;&#8364;33 saving per client sounds about right&#8221;) in ways they can&#8217;t validate large ones (&#8221;&#8364;2.4M in total savings&#8221; triggers skepticism, not confidence).</p><p><strong>Segment early, even if your segments are rough.</strong> A 25-person firm and a 350-person firm aren&#8217;t just different sizes, they&#8217;re different businesses with different economics. We defined three segments (Small, Medium, Large) with different assumptions for clients per user, usage patterns, and value capture rates. The segments were imperfect, but having them at all meant we could tailor conversations instead of forcing one story onto everyone.</p><p><strong>Build the model so the buyer can play with it.</strong> The single most effective sales tool we created wasn&#8217;t a pitch deck, it was a spreadsheet where the prospect could plug in their own numbers (number of users, clients per user, current cost per event) and see their specific ROI. When the buyer builds the business case themselves, they own it. That&#8217;s worth more than any pricing page.</p><p><strong>Your pricing is a communication tool, not just a revenue mechanism.</strong> The reason Version 1 failed wasn&#8217;t that the numbers were wrong. It was that the model couldn&#8217;t be understood in a meeting. Pricing needs to be explainable in one sentence. Ours is now: &#8220;You pay a percentage of the value we create per client, and you keep the rest.&#8221;</p><p><strong>Treat pricing as a moving target, not a launch decision.</strong> Demand curves shift as your brand strengthens, reference customers accumulate, and ROI evidence builds. We now revisit pricing quarterly and re-index annually (CPI plus a small premium). The teams that get pricing right aren&#8217;t the ones who pick a perfect number on day one. They&#8217;re the ones who build a feedback loop and use it.</p><h2>Common pricing mistakes worth avoiding</h2><p>A few patterns we keep seeing in other founders&#8217; pricing, often the same ones we made:</p><ul><li><p><em>Building a model that&#8217;s clever behind the scenes but opaque on the surface.</em> Complexity is fine internally. The buyer should still be able to explain what they&#8217;re paying for in one sentence.</p></li><li><p><em>Waiting too long to move to annual prepay.</em> Annual contracts smooth churn, generate cash, and force the renewal conversation once a year instead of twelve times. Push for it earlier than feels comfortable.</p></li><li><p><em>Treating pricing as static.</em> If you haven&#8217;t touched your pricing in two years, your pricing is wrong. Markets move, you move, your pricing has to move too.</p></li><li><p><em>Forgetting to embed concessions in the proposal.</em> Mid-market procurement teams will negotiate. If you don&#8217;t leave room, you&#8217;ll either lose the deal or lose margin you didn&#8217;t plan to lose.</p></li><li><p><em>Asking the wrong price discovery questions.</em> Don&#8217;t ask &#8220;what would you pay?&#8221; Ask &#8220;what are you currently paying for the alternative?&#8221; Relative beats absolute every time.</p></li></ul><h2>Your turn</h2><p>We&#8217;ve attached two resources below that you can use today:</p><p><strong>The Pricing Strategy Prompt</strong> &#8212; a structured prompt you can feed to any AI assistant along with your product details. It walks through the same value-based pricing framework we used: identifying your customer&#8217;s unit of value, modeling current vs. new cost, segmenting, and setting capture rates. It&#8217;s the thinking tool that got us from Version 1 to Version 4.</p><p><strong>The Pricing Model Template</strong> &#8212; a simplified version of the spreadsheet we use in sales meetings. Plug in your own segments, cost assumptions, and value capture rates. It calculates per-client value, per-user value, and total revenue potential automatically. Designed to be handed to a prospect so they can model their own ROI.</p><p>What pricing model are you using, and how did you land on it? We&#8217;re curious whether other founders went through the same iteration cycle or found a shortcut we missed. Drop a comment or reply, we read everything.</p><div><hr></div><h2>Resource 1: Pricing Strategy Prompt</h2><p><em>Copy this into Claude, ChatGPT, or any AI assistant along with a description of your product and customer.</em></p><p>I need help building a value-based pricing model for my product. Here&#8217;s the context:</p><p>[Describe your product: what it does, who uses it, what problem it solves]</p><p>Walk me through these steps:</p><p>1. UNIT OF VALUE: What is the smallest unit of value my customer experiences? (e.g., per transaction, per client served, per hour saved, per ticket resolved). Help me identify the right unit based on how my customer naturally thinks about their work.</p><p>2. CURRENT COST: Help me estimate what it costs my customer to handle one unit today WITHOUT my product. Break it down into: labor cost (hourly wage &#215; time), external/tool costs, and opportunity cost if relevant. Ask me clarifying questions to get realistic numbers.</p><p>3. NEW COST: Now estimate the cost per unit WITH my product. What time is saved? What external costs are eliminated? What&#8217;s the new per-unit cost?</p><p>4. VALUE CREATED: Calculate the value per unit (current cost minus new cost). Then help me think about how many units a typical customer handles per month/year.</p><p>5. CUSTOMER SEGMENTS: Help me define 2-3 customer segments (e.g., small/medium/large) based on volume, team size, or usage patterns. For each segment, estimate: units per user, number of users, total value created.</p><p>6. VALUE CAPTURE: What percentage of the value created should I capture as my price? Consider: competitive alternatives, customer willingness to pay, my cost to serve, and what leaves enough value on the table that the ROI is obvious.</p><p>7. TRIANGULATION CHECK: Compare the price you&#8217;ve derived against (a) the cost floor (what it costs me to serve one customer at scale, plus margin) and (b) the competitive anchor (what similar products charge). Does the value-based price sit comfortably above the floor and within reach of the anchor? If not, what needs to change?</p><p>8. SENSE CHECK: For each segment, calculate the implied monthly price per user. Does it feel reasonable? Can I explain it in one sentence? Would a buyer immediately understand the ROI?</p><p>9. PRICING STRUCTURE: Based on the above, recommend a pricing structure (per-user, per-unit, tiered, two-part tariff, or hybrid) and explain why it fits my market.</p><p>Present the final model as a simple table showing: segment, value per unit, units per user, value per user, capture rate, and price per user.</p><div><hr></div><h2>Resource 2: Pricing Model Template</h2><p><em>Below is the structure of our pricing model. Build this in a spreadsheet (or copy it into an AI and ask it to generate the formulas for you).</em></p><p><code>=== INPUTS (change these for each prospect) ===</code></p><p><code>Segment: [Small / Medium / Large]</code></p><p><code>Number of users: [e.g., 25 / 100 / 350]</code></p><p><code>Clients per user: [e.g., 50 / 100 / 200]</code></p><p><code>Hourly wage (blended): [e.g., &#8364;55]</code></p><p><code>--- Current process (without your product) ---</code></p><p><code>Time per event (hours): [e.g., 0.75]</code></p><p><code>External cost per event: [e.g., &#8364;5]</code></p><p><code>--- New process (with your product) ---</code></p><p><code>Time per event (hours): [e.g., 0.15]</code></p><p><code>External cost per event: [e.g., &#8364;2]</code></p><p><code>--- Your pricing ---</code></p><p><code>Value capture rate: [e.g., 20-30%]</code></p><p><code>=== CALCULATIONS ===</code></p><p><code>Current cost per client:</code></p><p><code> = (Hourly wage &#215; Time per event) + External cost</code></p><p><code> = (&#8364;55 &#215; 0.75) + &#8364;5 = &#8364;46.25</code></p><p><code>New cost per client:</code></p><p><code> = (Hourly wage &#215; New time) + New external cost</code></p><p><code> = (&#8364;55 &#215; 0.15) + &#8364;2 = &#8364;10.25</code></p><p><code>Value created per client:</code></p><p><code> = Current cost - New cost</code></p><p><code> = &#8364;46.25 - &#8364;10.25 = &#8364;36.00</code></p><p><code>Value per user (annual):</code></p><p><code> = Value per client &#215; Clients per user &#215; Events per year</code></p><p><code> = &#8364;36 &#215; 100 &#215; 1 = &#8364;3,600</code></p><p><code>Your price per user (annual):</code></p><p><code> = Value per user &#215; Capture rate</code></p><p><code> = &#8364;3,600 &#215; 25% = &#8364;900</code></p><p><code>Customer keeps per user:</code></p><p><code> = Value per user - Your price</code></p><p><code> = &#8364;3,600 - &#8364;900 = &#8364;2,700</code></p><p><code>ROI for customer:</code></p><p><code> = Customer keeps / Your price</code></p><p><code> = &#8364;2,700 / &#8364;900 = 3.0x</code></p><p><code>=== TOTAL REVENUE POTENTIAL ===</code></p><p><code>Revenue per segment:</code></p><p><code> = Price per user &#215; Number of users</code></p><p><code> = &#8364;900 &#215; 100 = &#8364;90,000/year</code></p><p><code>Your cost to serve: [your platform cost per user]</code></p><p><code>Gross margin per user: [price minus cost to serve]</code></p><p><em>Pro tip: build this as a live spreadsheet and hand it to your prospect during the sales meeting. Let them change the inputs. When they build the business case themselves, they sell it internally for you.</em></p><p><em>Image credit: Matheus Bertelli via Pexels</em></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading! Subscribe for free to receive new posts and support my work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div>]]></content:encoded></item><item><title><![CDATA[Write the Memo First]]></title><description><![CDATA[Why your investment memo is the first thing, not the last]]></description><link>https://blueishprint.thirdvector.io/p/write-the-memo-first</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/write-the-memo-first</guid><dc:creator><![CDATA[Friso Paping]]></dc:creator><pubDate>Wed, 22 Apr 2026 08:31:17 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!zwDu!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!zwDu!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!zwDu!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg 424w, https://substackcdn.com/image/fetch/$s_!zwDu!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg 848w, https://substackcdn.com/image/fetch/$s_!zwDu!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!zwDu!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!zwDu!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg" width="1456" height="971" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:971,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:558058,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/jpeg&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://thirdv3ctor.substack.com/i/195008250?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!zwDu!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg 424w, https://substackcdn.com/image/fetch/$s_!zwDu!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg 848w, https://substackcdn.com/image/fetch/$s_!zwDu!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!zwDu!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7b84218b-2cbf-45f6-a7d7-d7f32a17df79_5472x3648.jpeg 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p></p><p>Doesn&#8217;t matter if you&#8217;re building a consumer product, B2B software, or sending rockets to Mars. The first thing we write when starting a new business or project is an investment memo. Not a plan. Not a pitch deck. 5 to 8 pages, for an audience of nobody, on what we think the problem is, the target customer, why now, the solution, why you are the one to solve it, and how to get it into the hands of paying customers.</p><p>This sounds obvious until you notice that almost nobody starts here. The usual order is the opposite. Open a spreadsheet, scrape an ICP list, run the TAM math, and only after a few weeks of that try to articulate what you actually believe. You end up a month deep in research about a problem you never properly defined.</p><p>This matters more now than it used to, because of AI. Used to be that getting good information took real time. Now you can run a deep research report in half an hour. You can have a competitor map by lunch. The constraint has moved. It used to be the research itself. Now it&#8217;s knowing what to point the research at, and that part still has to come from a human writing down what they believe.</p><h2><strong>What this is</strong></h2><p>We spent this weekend building a Claude Skill that walks a founder through writing an Investment Memo for a new business. The memo is a 5 to 8 page Word document, written in full prose. We tested, iterated, and used the skill on ourselves.</p><p>The skill is open source and sits here: <a href="https://github.com/third-vector-eu/claude-skills/tree/main/investment-memo">github.com/third-vector-eu/claude-skills</a>.</p><h2><strong>The process</strong></h2><p>The skill runs five phases. You don&#8217;t need to memorise them, but knowing what&#8217;s coming helps.</p><ol><li><p><strong>Calibrate.</strong> It asks how familiar you are with the domain. Are you a domain expert who has lived in this market for a decade? Adjacent, with related but not identical experience? Or exploring new territory? The answer changes how the rest of the interview runs. Domain experts get harder questions. Explorers get more research support and more humility prompts.</p></li><li><p><strong>Interview.</strong>Section by section, the skill walks you through the twelve parts of the memo. Problem, target customer, why now, solution, founder-market fit, business model, go-to-market, competition, customer discovery, risks and falsification, milestones, the ask. For each section it asks three or four sharp questions and runs parallel web research on market signals, competitors, and &#8220;why now&#8221; evidence. Every answer gets reflected back before moving on.</p></li><li><p><strong>Draft.</strong> It writes the full memo in prose. Every major claim gets tagged either <em>[Conviction]</em> or <em>[Hypothesis]</em>, so you and any future reader know which parts are earned and which parts are bets.</p></li><li><p><strong>Critical review.</strong> This is where the skill earns its keep. It runs the draft against a fourteen-point checklist built around the specific failure modes of early-stage memos. Fluff that looks like content. Contradictions between sections. Weak claims. Missing evidence. Confidence outrunning reality. It then asks you the three sharpest questions, the ones you should sit with this week.</p></li><li><p><strong>Iterate.</strong> You answer the critiques. The memo gets revised. A companion document captures the open questions and next experiments that fell out of the review, which becomes the punch list for your next 30 days.</p></li></ol><p>The full process takes between 1.5 and 3 hours, depending on how deeply you think through each answer. Sit with the questions. The slower answers are usually the better ones.</p><h2><strong>Why write this yourself</strong></h2><p>You cannot outsource founder conviction. You cannot outsource the felt sense that you are a credible person, or at least a plausible one, to build this company. The act of writing the memo <em>is</em>the thinking. Someone else writing it for you means you still have to do the thinking later, except now you have a document that sounds like you didn&#8217;t.</p><p>There is a second, quieter reason. The memo is the spine of everything downstream. The pitch deck is an extract of the memo. The website copy is an extract of the memo. The sales messaging, the hiring pitch, the first three blog posts, the cold email to a design partner, the opening talk at your first conference. All of it references the memo. If the memo is weak, every artefact downstream is weak. If the memo is sharp, every artefact downstream inherits the sharpness.</p><p>Write it once. Properly. Then the next month&#8217;s work compounds off it.</p><h2><strong>It is a living document</strong></h2><p>The first draft is not the final draft.</p><p>You will have ten conversations with prospects and realise your ICP is wrong. You will price something and watch three buyers flinch at the number. You will discover a competitor you had missed. You will learn that the pain point you thought was most acute is actually second, and the real urgency sits elsewhere.</p><p>Come back to the skill. Run it again. Revise the memo. What changed between v1 and v2 is where your learning actually lives. Keep the old versions. The delta is the story of your company.</p><h2><strong>Our tip: read it out loud</strong></h2><p>When the draft is done, read it out loud. To your co-founder. To your partner. To a patient friend.</p><p>Reading it out loud does two things at once. You catch yourself hedging. You hear the sentences where you lose conviction halfway through and start padding. Your voice gets softer. You speed up. You know, as you are reading, which paragraphs do not earn their space.</p><p>At the same time, the people listening start asking questions. And the good ones really hurt. The questions you haven&#8217;t thought about yet. The questions you were hoping no one would ask. Those are the ones you write down, and the answers you go find this week.</p><h2><strong>Your turn</strong></h2><p>The skill is on GitHub: <a href="https://github.com/third-vector-eu/claude-skills/tree/main/investment-memo">github.com/third-vector-eu/claude-skills</a>.</p><p>Write yours. Read it out loud. Tell us where it breaks.</p>]]></content:encoded></item><item><title><![CDATA[Third Vector Observer | Edition 3]]></title><description><![CDATA[Agents Need Context, Anthropic's Growth, SAP Buys Reltio, 1300+ Marketing Skills]]></description><link>https://blueishprint.thirdvector.io/p/third-vector-observer-edition-3</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/third-vector-observer-edition-3</guid><dc:creator><![CDATA[Third Vector]]></dc:creator><pubDate>Fri, 17 Apr 2026 14:29:12 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!bMgj!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!bMgj!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!bMgj!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg 424w, https://substackcdn.com/image/fetch/$s_!bMgj!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg 848w, https://substackcdn.com/image/fetch/$s_!bMgj!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!bMgj!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!bMgj!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg" width="1456" height="971" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:971,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:1428539,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/jpeg&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://thirdv3ctor.substack.com/i/194522263?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!bMgj!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg 424w, https://substackcdn.com/image/fetch/$s_!bMgj!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg 848w, https://substackcdn.com/image/fetch/$s_!bMgj!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg 1272w, https://substackcdn.com/image/fetch/$s_!bMgj!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F77d02d4c-b3ff-4247-b394-04cb9caaa0a0_4857x3238.jpeg 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><h1>Third Vector Observer | Edition 3</h1><p><em>Welcome to the Third Vector Observer, an easy-to-digest newsletter featuring 3-5 high-quality, insightful pieces on agentic AI and the impact on organizations and markets. Delivered to your inbox roughly once a month.</em></p><h2>Things We Observed...</h2><h3>1. Your Data Agents Need Context (and a Semantic Layer Won&#8217;t Cut It)</h3><p>a16z published a piece that crystallizes something we think is fundamental: data agents keep failing because nobody gave them the business context they need to do anything useful.</p><p>The example is deceptively simple. Ask a data agent &#8220;what was revenue growth last quarter?&#8221; and watch it drown. Which revenue definition? ARR or run rate? Which fiscal quarter? Which table is the source of truth when finance uses fct_revenue but the data team built mv_revenue_monthly? The model can write flawless SQL and still return garbage.</p><h4>Semantic Layer vs. Context Layer vs. Ontology</h4><p>These three concepts keep getting conflated, but they&#8217;re distinct:</p><ul><li><p><strong>Semantic layer</strong> (LookML, dbt metrics): Defines specific metrics in code. Revenue equals X, churn equals Y. Useful but narrow, mapping well to BI dashboards where queries are predictable.</p></li><li><p><strong>Context layer</strong>: Broader. Encompasses metric definitions but adds entity relationships, identity resolution, tribal knowledge, governance rules, and workflow-specific instructions. Think of it as the difference between giving someone a glossary versus six months of institutional knowledge.</p></li><li><p><strong>Ontology</strong>: Formally defines how entities relate to each other. A customer <em>has</em> accounts, accounts <em>have</em> subscriptions, subscriptions <em>generate</em> revenue events. Palantir built a massive business on exactly this idea.</p></li></ul><p>The ontology is the structural backbone; the context layer is the intelligence that makes it operational for agents.</p><h4>Where the Market Is Heading</h4><p>a16z maps out three categories of emerging solutions:</p><ul><li><p><strong>Data gravity platforms</strong> (Databricks, Snowflake) adding lightweight context features</p></li><li><p><strong>Existing AI analyst companies</strong> pivoting to context-first approaches</p></li><li><p><strong>Dedicated context layer startups</strong> building from scratch</p></li></ul><p>Encoding every business rule into YAML files by hand won&#8217;t scale. We believe the winners will crack automated context construction with human refinement.</p><p><a href="https://www.a16z.news/p/your-data-agents-need-context">Your Data Agents Need Context</a></p><div><hr></div><h3>2. Anthropic&#8217;s Revenue Growth Is Breaking the Playbook</h3><p>Full disclosure: we at Third Vector are massive fans of Claude and have integrated Claude Cowork and Claude Code into how we work. So we&#8217;re biased. But the numbers <strong>Tomasz Tunguz</strong> laid out this week are staggering on their own.</p><p><strong>Anthropic added $10 billion in revenue in a single month.</strong> That&#8217;s twice Databricks&#8217; annual run rate, absorbed in thirty days.</p><h4>The Trajectory in Perspective</h4><ul><li><p><strong>ServiceNow:</strong> 20 years to hit $10B annual revenue</p></li><li><p><strong>Shopify:</strong> 18 years</p></li><li><p><strong>Palo Alto Networks:</strong> 19 years</p></li><li><p><strong>Anthropic:</strong> 42 months</p></li></ul><h4>Can Anthropic Surpass NVIDIA?</h4><p>Tunguz&#8217;s core question is <em>when</em>, not <em>whether</em>. NVIDIA currently generates <strong>$215B</strong> in annual revenue at a 22x multiple, producing a $4.8 trillion market cap. At a 25x forward multiple, Anthropic would need roughly $200B in annual revenue to match. The bull case says three years. The base case says four. Even the bear case gets there in seven.</p><p>The critical caveat is <strong>customer concentration risk</strong>. But the sheer velocity is unlike anything the software industry has produced.</p><p>What strikes us is what this means for the broader ecosystem. When an AI model company grows this fast, every adjacent category gets pulled forward: infrastructure, tooling, implementation services, data preparation. The companies building agentic workflows on top of these models are riding actual exponential demand.</p><p><a href="https://www.tomtunguz.com/anthropic-most-valuable-company/">When Will Anthropic Surpass NVIDIA?</a></p><div><hr></div><h3>3. SAP Buys Reltio: Do We See A Pattern Emerging?</h3><p>Salesforce acquired Informatica. Now SAP is buying Reltio. Two of the largest enterprise software companies on the planet looked at their agentic AI ambitions and came to the same conclusion: <strong>the data underneath is a mess, and they can&#8217;t fix it themselves.</strong></p><h4>What Reltio Does</h4><p>Reltio is a cloud-native <strong>master data management (MDM)</strong> platform that creates a single, consistent view of customers, products, suppliers, and employees across both SAP <em>and</em> non-SAP systems. That last part matters.</p><p>SAP&#8217;s head of product engineering, Muhammad Alam, was quoted saying:</p><blockquote><p>&#8220;AI cannot reach its full potential when data is fragmented across business units, platforms and domains.&#8221;</p></blockquote><p>In other words: if your agents can&#8217;t trust the data, they can&#8217;t do anything useful. SAP launched its Business Data Cloud a year ago, but building clean data infrastructure from scratch is slow. Buying Reltio gives them entity resolution and data harmonization that would have taken years to build internally.</p><h4>The Bigger Signal</h4><p>ERP companies are waking up to the fact that decades of bolt-on acquisitions and customization have left enterprise data <strong>fundamentally incompatible with autonomous AI workflows.</strong> Agentic systems need consistent entities, resolved identities, and governed data. That&#8217;s precisely what MDM companies have been building for years, mostly as unsexy infrastructure nobody wanted to pay for. Now it&#8217;s suddenly strategic.</p><p><a href="https://cloudwars.com/innovation-leadership/sap-pumps-up-agentic-ai-and-data-chops-with-reltio-acquisition/">SAP Pumps Up Agentic AI and Data Chops with Reltio Acquisition</a></p><div><hr></div><h3>4. Claude Managed Agents: The Infrastructure Layer Catches Up</h3><p>Until now, building production-grade AI agents meant stitching together your own sandboxing, authentication, error recovery, and orchestration. Months of plumbing before a single line of agent logic. <strong>Anthropic just collapsed that timeline.</strong></p><p>Claude Managed Agents, launched April 8, is a suite of composable APIs that handles the operational complexity of deploying cloud-hosted agents at scale:</p><ul><li><p><strong>Secure sandboxing</strong> for code execution</p></li><li><p><strong>Built-in credential management</strong> with scoped permissions</p></li><li><p><strong>Persistent checkpointing</strong> that survives disconnections</p></li><li><p><strong>End-to-end tracing</strong> so you can inspect every tool call and decision point</p></li></ul><p>In internal testing, the platform improved task success rates by up to <strong>10 points</strong> over standard prompting loops, with the biggest gains on complex, multi-step tasks.</p><h4>Enterprise Adoption Signals</h4><p>The early deployments are production, not proofs of concept:</p><ul><li><p><strong>Notion</strong> is using it for parallel task execution agents</p></li><li><p><strong>Rakuten</strong> deployed agents across product, sales, marketing, and finance within weeks</p></li><li><p><strong>Sentry</strong> built a debugging agent that diagnoses issues <em>and</em> writes patches</p></li></ul><p>Also notable: the <strong>multi-agent coordination</strong> capability (in research preview) allows agents to delegate subtasks to other agents in parallel. That&#8217;s the architecture pattern everyone&#8217;s been talking about, now available as a managed service.</p><h4>What This Changes</h4><p>The build-versus-buy calculus just shifted. Security and compliance overhead alone used to be a six-month project. When the model provider handles sandboxing, identity, and tracing natively, the barrier to shipping drops to the quality of your prompts and tool definitions. We&#8217;re moving from &#8220;can we build this?&#8221; to &#8220;how fast can we ship it?&#8221;</p><p><a href="https://claude.com/blog/claude-managed-agents">Claude Managed Agents: Get to Production 10x Faster</a></p><div><hr></div><h3>5. 1,300 Marketing Skills Walk Into an AI Agent</h3><p>Buron just dropped something that caught our eye: a library of <strong>over 1,300 pre-built marketing skills</strong> designed to plug directly into Claude and other AI agents. Each skill is a structured knowledge guide covering a specific marketing task, from diagnosing landing page conversion problems to running SEO audits optimized for AI engine discovery.</p><h4>What&#8217;s in the Box</h4><p>The skills span six categories:</p><ul><li><p><strong>SEO</strong>, including audits, schema markup, E-E-A-T authority building</p></li><li><p><strong>Paid Ads</strong>, including Google Ads account analysis and conversion tracking</p></li><li><p><strong>Content Marketing</strong>, including creation, repurposing, optimization</p></li><li><p><strong>Strategy</strong>, including campaign planning and competitive analysis</p></li><li><p><strong>Analytics</strong>, including performance diagnostics and attribution</p></li><li><p><strong>Social Media</strong>, including campaign planning and execution</p></li></ul><p>There&#8217;s a community leaderboard where skills get rated, which creates a natural quality filter. The top-rated ones read more like specialist playbooks than simple prompt templates.</p><h4>Why This Matters Beyond Marketing</h4><p>We&#8217;ve been building custom skills at Third Vector, and the craft involved in writing a genuinely useful skill is non-trivial. You need to encode domain expertise, edge cases, and decision logic. Buron is betting that a marketplace approach can scale this faster than every company building bespoke.</p><p><a href="https://www.marketingskills.sh/">Buron Marketing Skills</a></p>]]></content:encoded></item><item><title><![CDATA[How to build go-to-market from scratch]]></title><description><![CDATA[In this piece, we walk through the full process, from foundation to expansion, and explain why the order matters.]]></description><link>https://blueishprint.thirdvector.io/p/how-to-build-go-to-market-from-scratch</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/how-to-build-go-to-market-from-scratch</guid><dc:creator><![CDATA[Friso Paping]]></dc:creator><pubDate>Tue, 14 Apr 2026 18:11:44 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/dcc5d056-ad11-4f10-97b1-baab0a2fd0c9_4500x3000.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<h2><strong>TL;DR</strong></h2><p>Building a go-to-market is not a single strategy document. It&#8217;s a sequence of phases, each with a gate you need to pass before moving on.</p><ul><li><p><strong>Phase 0 (Setup):</strong> Get your file storage, AI tools, and foundational skills configured before you do anything else.</p></li><li><p><strong>Phase 1 (Foundation):</strong>Write your investment memo, map the product value, size the market. If TAM/SAM doesn&#8217;t hold up, go back.</p></li><li><p><strong>Phase 2 (Validate):</strong>Define your ICP, build a target list, run experiments. If the messaging doesn&#8217;t land, refine and repeat.</p></li><li><p><strong>Phase 3 (Scale):</strong> Build the machine. Content, sales process, CRM, inbound, pipeline generation, pricing, KPIs, onboarding, win/loss analysis. Gate: repeatable wins and consistent pipeline velocity.</p></li><li><p><strong>Phase 4 (Expand):</strong> Grow what you have. Feature release process, upsell/cross-sell, pricing review.</p></li><li><p><strong>Strategy (parallel):</strong> Runs alongside Phase 3 and 4. Full TAM/SAM deep-dive, unit economics, budget planning, forecasting, annual GTM strategy.</p></li></ul><p>Every step comes with deliverables: AI skills, guides, prompt-templates, and tools. And six processes never stop running: your investment memo, ICP definitions, product experimentation, pricing, positioning &amp; messaging, and customer feedback.</p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading! Subscribe for free to receive new posts and follow our work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p><h2><strong>The full process at a glance</strong></h2><p>We built an interactive flowchart that maps out the entire go-to-market process. All 25 steps, three gates, the strategy track, and the deliverables for each step. You can toggle between a high-level overview and the detailed version.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!7D-0!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!7D-0!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png 424w, https://substackcdn.com/image/fetch/$s_!7D-0!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png 848w, https://substackcdn.com/image/fetch/$s_!7D-0!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png 1272w, https://substackcdn.com/image/fetch/$s_!7D-0!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!7D-0!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png" width="1456" height="2128" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:2128,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:null,&quot;alt&quot;:&quot;The Go-to-Market Process flowchart showing phases 0 through 4, gates, strategy track, and always-running processes&quot;,&quot;title&quot;:null,&quot;type&quot;:null,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:null,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="The Go-to-Market Process flowchart showing phases 0 through 4, gates, strategy track, and always-running processes" title="The Go-to-Market Process flowchart showing phases 0 through 4, gates, strategy track, and always-running processes" srcset="https://substackcdn.com/image/fetch/$s_!7D-0!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png 424w, https://substackcdn.com/image/fetch/$s_!7D-0!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png 848w, https://substackcdn.com/image/fetch/$s_!7D-0!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png 1272w, https://substackcdn.com/image/fetch/$s_!7D-0!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8d2bbc7e-441f-4d54-bf77-24759c23162c_2200x3216.png 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p><a href="https://www.thirdvector.io/guides/gtm-flowchart.html">View the interactive GTM flowchart &#8594;</a></p><p>The rest of this article walks through each phase and explains why they&#8217;re in this order.</p><h2><strong>Why order matters</strong></h2><p>Most go-to-market advice skips the sequencing. You get a list of things you should do, tactics that worked for someone else, a framework with four quadrants. What&#8217;s missing is the order, and the gates that tell you whether you&#8217;re ready for the next step.</p><p>We recommend a structured, gated go-to-market process that starts from first principles and scales to a full annual strategy. We split it into five phases, each with clear steps, and each ending with a checkpoint. If you pass the gate, you move on. If you don&#8217;t, you go back and fix what&#8217;s broken.</p><p>For every step in the process, we provide four types of deliverables. Skills are specialized instructions that teach AI models specific workflows, styles, or domain expertise, allowing them to perform complex, repetitive tasks efficiently. Guides are deep-dive articles with practical how-tos or frameworks for the team. Prompt-templates are structured prompts you configure with your own context and run tasks. And tools are software recommendations for the infrastructure you need.</p><p>Let&#8217;s walk through it.</p><p></p><h2><strong>Phase 0: Setup</strong></h2><p>Before you build anything, make sure your basics are in place. This sounds obvious, but we see teams jump into outreach or content creation with no shared file structure, no AI tooling configured, and no shared language around what they&#8217;re building.</p><p>Three things to sort out first. Set up your shared file storage and connect the AI tools you&#8217;ll use throughout the process. Configure your foundational AI skills (such as the <a href="https://github.com/blader/humanizer/blob/main/SKILL.md">humanizer</a>, so your output doesn&#8217;t read like a chatbot wrote it), tone of voice, formatting standards. And feed your AI the right context: industry keywords, jargon, a high-level description of what you&#8217;re building. Context in, quality out.</p><p>This takes a day. Maybe two. But it saves you weeks of rework later.</p><p></p><h2><strong>Phase 1: Foundation</strong></h2><p><em>&#8220;What are we selling, to whom, and is the market big enough?&#8221;</em></p><p>You start with an investment memo. This is a founder exercise. You write down the value, the product, the target market in a short, crisp document. Think of it as an internal pitch, something anyone on the team can read and immediately understand what you&#8217;re doing and why.</p><p>From there, you do product and value mapping. What do we aspire the product to be? What value do we expect it to deliver? This matters especially when you&#8217;re not building a mass consumer product. In B2B, thinking explicitly about value delivery is the foundation of everything that follows, from positioning to pricing to how you structure your sales conversations.</p><p>Then you size the market. Back of the envelope only. You can kill yourself on TAM/SAM analysis, and many teams do. At this stage, you just need to know: is this market big enough to be worth pursuing? A quick top-down and bottom-up estimate. If the answer is no, you go back to step one and rethink your product or your target.</p><p>That&#8217;s Gate 1. Pass it, and you continue.</p><p>After the gate, you do a quick competitive landscape scan. Who else is in this market? How do they position themselves? What problems do they claim to solve? This is not a detailed battle card exercise. That comes later. This is orientation.</p><p>Then revenue target math. You probably have an ARR or MRR goal. Work backwards. How many logos do you need? What deal size does that imply? What kind of sales motion can realistically support that volume? If you need to sign 200 enterprise clients in your first year without a sales team, that number is telling you something. Either your target is unrealistic, your pricing needs to change, or your go-to-market motion needs to be fundamentally different.</p><p>The phase ends with an MVP definition. What are we building? Who are we selling to? What are the minimum product requirements to test our assumptions?</p><p></p><h2><strong>Phase 2: Validate</strong></h2><p><em>&#8220;Does our value land with the right people?&#8221;</em></p><p>You start with ICP definition. Ideal Customer Profile. This is a team exercise, not a solo founder activity. Get everyone who touches customers into a room, customer support, implementation, tech, sales, and define what your ideal client looks like. Crucially, define it using data points you can actually measure externally. &#8220;Companies that care about innovation&#8221; is not an ICP. &#8220;B2B SaaS companies with 50-200 employees, Series A or later, with a customer support team of 10+&#8221; is.</p><p>ICP will evolve over the lifetime of your company. It&#8217;s never fixed. But you need a starting point to test against.</p><p>Once you have your ICPs, you work on positioning and messaging. How do you speak to each type of ICP? What&#8217;s the persona within those companies? What problems do they lose sleep over? This is where a lot of teams skip ahead to outreach without doing the positioning work first, and then wonder why their cold emails get ignored.</p><p>Next, build your initial target list. Combine your ICP definition with externally measurable data points to populate a list of companies. Then find the right individuals within those companies. If you&#8217;re going after a consumer market, the company layer drops away. But for B2B, it&#8217;s always two steps: which company, then which person.</p><p>Now you design experiments. You&#8217;re going to run multiple outreach experiments in parallel, with short review cycles, testing different variables: messaging, channels, cadences, value propositions. Each experiment targets a specific ICP group. Minimum two, ideally four to six. The goal is to learn which combination of ICP, message, and channel actually converts.</p><p>Before you run those experiments, you need a pricing hypothesis. Prospects will ask about pricing early. If you don&#8217;t have a ballpark, you&#8217;ll get checked out of conversations fast. There are three approaches: cost-based, value-based, and competitive-based pricing. At this stage, you pick one direction and test it. The deep pricing work comes later.</p><p>Then you run aggressive outreach cadences, measure results, and collect feedback. What are the numbers telling you? What are prospects actually saying?</p><p>Gate 2: do you see emerging patterns? Do you understand how to reach your targets consistently? If yes, you move to scale. If not, you loop back to ICP definition and refine.</p><p></p><h2><strong>Phase 3: Scale</strong></h2><p><em>&#8220;How do we build the machine?&#8221;</em></p><p>Content comes first. Everything you learned in Phase 2, the messaging that worked, the objections you heard, the language your ICPs use, feeds directly into your content. A lot of content creation can be accelerated with AI now, but the quality of that content depends entirely on the quality of the input.</p><p>In parallel, you build three things: your sales process (how deals actually move through stages), your CRM and go-to-market infrastructure (the systems that capture everything), and your inbound engine (website, SEO, demos, content distribution). These three are interdependent and should be built together.</p><p>Once that infrastructure is in place, you systematize pipeline generation. Outbound and inbound, measured continuously. Pipeline measurement is where many teams start to feel real about their go-to-market. You&#8217;re no longer guessing, you&#8217;re watching deals move.</p><p>As deals progress through the pipeline, pricing becomes critical. This is where you move from hypothesis to real pricing work. Value-based, cost-based, or competitive, it doesn&#8217;t matter which framework, what matters is that you approach it as a discipline, not an art. We have dedicated guides and templates for this.</p><p>Define your metrics and KPIs. When you set up your CRM and sales process, you already started thinking about what to measure. But once pipeline starts flowing, you need to know how to act on those metrics. What we see a lot is that pipelines are based on hope rather than data. Conversion rates, pipeline velocity, deal cycle times, these are the numbers that tell you whether your machine is actually working.</p><p>In a separate workstream, you think about onboarding and first value delivery. If this breaks, pipeline doesn&#8217;t matter. How fast can you get customers live and delivering the value you promised? And you run win/loss analysis on every closed deal. Why did we win? Why did we lose? Link the answers back to your ICP definition.</p><p>Gate 3: repeatable wins and consistent pipeline velocity. Deals move through the pipeline at a predictable rate. Conversion rates are steady month over month. You can have different motions running side by side, different countries, different products, different segments. But each motion needs to be repeatable on its own.</p><p></p><h2><strong>Phase 4: Expand</strong></h2><p><em>&#8220;How do we grow what we have?&#8221;</em></p><p>Once you start onboarding customers, the cheapest revenue is sitting right in front of you. It&#8217;s always easier to sell to an existing customer where you&#8217;ve already done the legal work and built the relationship.</p><p>This phase covers three things. First, your feature release process. How do new features reach existing clients? Think release cadence, product marketing for existing customers, and pricing implications of new functionality. Second, your upsell and cross-sell strategy. What triggers an expansion conversation? What upgrade paths exist? How does the commercial team identify and act on those signals? Third, a pricing review against actual usage data, expansion revenue, and competitive moves.</p><p></p><h2><strong>The strategy track</strong></h2><p>Strategy runs in parallel with Phase 3 and 4. It doesn&#8217;t wait until you&#8217;ve built the machine. Once you have your foundation and your initial ICP validated, you should start planning ahead.</p><p>We always recommend planning at least a year out. It forces you to step back from the day-to-day. What do we want revenue to be? What can we invest to get there? How big does the team need to be, and what performance do they need to deliver? What&#8217;s the average deal size we need to sign?</p><p>The strategy track includes a full TAM/SAM analysis overlaid against your actual pipeline penetration. Unit economics validation, are we reaching our goals cost-effectively? Budget and resource planning, including headcount, ramp time, and performance targets. And forecasting, which used to be more art than science but doesn&#8217;t have to be. You can triangulate cohort analysis, conversion rates, bottom-up enterprise modeling, and velocity-adjusted weighted pipeline to build a forecast you can actually trust.</p><p>All of this feeds into your annual go-to-market strategy, reviewed quarterly, refreshed annually. If your business depends on year-end deals or chunky enterprise contracts, plan the timing of that strategy cycle carefully. You want your go-to-market plan ready before the new financial year starts.</p><p></p><h2><strong>What never stops</strong></h2><p>Six processes run continuously from the moment they start:</p><ol><li><p>Your <em>investment memo </em>updates as you learn what works. It&#8217;s your elevator pitch for investors, for new hires, for anyone who needs to understand what you sell and why. Keep it current.</p></li><li><p>Your <em>ICP definitions</em> evolve as your product grows and your markets expand.</p></li><li><p><em>Product experimentation</em> continues to validate whether product value and ICP align, ideally measured live within the product.</p></li><li><p><em>Pricing</em> refines as you add functionality and close more deals.</p></li><li><p><em>Positioning and messaging</em> sharpens with every customer conversation.</p></li><li><p><em>Customer feedback</em> feeds back into everything, ICP definitions, product marketing, feature prioritization.</p></li></ol><p></p><h2><strong>What&#8217;s next</strong></h2><p>In future Blueishprint articles, we&#8217;ll go deep on individual steps. Some will get their own article, others we&#8217;ll combine. We&#8217;ll share the skills, guides, prompt-templates, and tools for each step as we go.</p><p>Next up: the <strong>Investment Memo</strong>. How to write one, what it should contain, and why it&#8217;s the single document everyone on your team should be able to recite.</p><p><em>Factory image: Alan Wang (Pexels)</em></p><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading! Subscribe for free to receive new posts and follow our work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div>]]></content:encoded></item><item><title><![CDATA[We're Building in Public. Here's Everything.]]></title><description><![CDATA[Two brothers-in-law, a handful of companies, one acquisition, and a growing suspicion we keep solving the same problems.]]></description><link>https://blueishprint.thirdvector.io/p/were-building-in-public-heres-everything</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/were-building-in-public-heres-everything</guid><dc:creator><![CDATA[Friso Paping]]></dc:creator><pubDate>Tue, 07 Apr 2026 19:21:36 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!bmLu!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p><em>Written by: <span class="mention-wrap" data-attrs="{&quot;name&quot;:&quot;Friso Paping&quot;,&quot;id&quot;:207664472,&quot;type&quot;:&quot;user&quot;,&quot;url&quot;:null,&quot;photo_url&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/6a776d71-90e4-4c0f-8eaa-c86a46942b50_800x800.jpeg&quot;,&quot;uuid&quot;:&quot;8f32c9b8-418d-45bf-8732-2cbc74ff1cd8&quot;}" data-component-name="MentionToDOM"></span> &amp; <span class="mention-wrap" data-attrs="{&quot;name&quot;:&quot;Tim de Rooij&quot;,&quot;id&quot;:17909807,&quot;type&quot;:&quot;user&quot;,&quot;url&quot;:null,&quot;photo_url&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/96d97886-525e-42ec-b171-b33c54924773_608x608.png&quot;,&quot;uuid&quot;:&quot;93d6bbb1-6274-4b22-ae5a-91933c02eb4b&quot;}" data-component-name="MentionToDOM"></span> </em></p><p>One of us is writing this with a product launch in three weeks that he&#8217;s not ready for. A digital identity vault going live at a public event. The plan is to launch &#8220;The Founding Hundred,&#8221; exclusive access for the first hundred people to sign up at &#8364;100 per month. A golden badge, early access, and the kind of FOMO we&#8217;re betting will get people moving. If we nail it, that&#8217;s &#8364;120K in annual revenue from day one. The product is still in progress. The website isn&#8217;t live yet. There&#8217;s no final decision on the consumer brand name. Scary? Sure. But we&#8217;ll take it as it comes.</p><p style="text-align: justify;">That&#8217;s what building a business actually looks like. Not the polished version. The real one.</p><p style="text-align: justify;">A few months ago, one of us called the other while setting up a new holding structure. &#8220;What accountant did you use for the BV again?&#8221; The same question had gone the other direction the week before, except that time it was about pricing tiers. The week before that, it was GTM sequencing. Different businesses, different markets, and yet every single week we were having the exact same conversation: &#8220;How did we do this last time?&#8221;</p><p style="text-align: justify;">At some point the response was: &#8220;We need to stop calling each other and start writing this down.&#8221; And the reply: &#8220;Yeah, but properly this time.&#8221; That became <em>The Blueishprint</em>.</p><h2 style="text-align: justify;"><strong>Who we are</strong></h2><p style="text-align: justify;">Tim is 39. Friso is 33. We&#8217;re brothers-in-law, which means we didn&#8217;t choose to be family but we did choose to keep talking about business at every family dinner until everyone else left the table.</p><p style="text-align: justify;">We&#8217;re not co-founders. We&#8217;re not building one thing together. We&#8217;re each starting our own businesses, in different markets, with different products. But our backgrounds are different enough that we keep filling in each other&#8217;s blind spots.</p><p style="text-align: justify;">Tim has lived and worked across the US, Europe, and Asia. He&#8217;s built teams and delivered AI products across all of them and has this annoying ability to look at a business model and immediately spot the thing that doesn&#8217;t add up. Right now he&#8217;s building Third Vector, helping companies redesign their operations around agentic AI. Not adding tools to existing workflows, but rethinking how the work gets done.</p><p style="text-align: justify;">Friso built Sentinels, an AI-powered transaction monitoring platform, and sold it to a strategic buyer. Eight years in total, from first pitch deck to acquisition integration and managing the post-acquisition phase. The product is now running at some of the largest financial institutions in the world. It was an incredibly fun ride with plenty of ups, downs, and more near-death experiences than we&#8217;d like to admit. The acquisition alone taught us more in six months of due diligence than in years of actually running the business.</p><p style="text-align: justify;">Now Friso is building a digital identity vault. You store your personal data once, and when an institution needs something, you approve the request and the data flows over, verified and up to date. No more digging up passport scans. We&#8217;ll write a lot more about this one.</p><p style="text-align: justify;">Between us we&#8217;ve started companies, raised tens of millions in funding, held roles at organizations we probably had no business being at, and survived enough mistakes to fill a very long spreadsheet. We both love building things. That&#8217;s the short version.</p><h2 style="text-align: justify;"><strong>Introducing </strong><em><strong>The Blueishprint</strong></em></h2><p style="text-align: justify;">We&#8217;re both moving about ten times faster this time around. Partly because we&#8217;ve been here before and know which shortcuts are real and which ones will blow up in your face. Partly because of AI. The combination of experience and the tools available right now means we can build in months what used to take years.</p><p style="text-align: justify;">So why keep all of this in our heads?</p><p style="text-align: justify;">Most business content is either too abstract (&#8221;here are 7 principles of great leadership&#8221;) or too specific to someone else&#8217;s situation to be useful for yours. What actually helped us came from people who shared the messy middle. Not polished case studies from five years later, but real decisions, made in real time, before anyone knew how they&#8217;d turn out.</p><p style="text-align: justify;">That&#8217;s <em>The</em> <em>Blueishprint</em>, which we&#8217;ll be publishing on Third Vector&#8217;s Substack.</p><p style="text-align: justify;">Blue-ish-print. Because a blueprint implies you&#8217;ve got it all figured out. We don&#8217;t, and in the age of AI things are changing so fast that what we write today may become obsolete tomorrow.</p><p style="text-align: justify;">And we won&#8217;t be the only ones writing. We&#8217;ll bring in co-authors who are building right now and have something specific to share. Every article comes with actual materials. The spreadsheet you can fill in that afternoon. The workshop agenda you can run with your team on Friday. We&#8217;ll share the actual tools we used. Prompts, Lucid charts, Claude skills, case studies. Not because we think our way is the only way, but because reading about a GTM strategy is useless if you can&#8217;t build one yourself the same afternoon.</p><h2 style="text-align: justify;"><strong>The tightrope: product and GTM</strong></h2><p style="text-align: justify;">Every time either of us starts a new business or project, the approach is roughly the same. You start with the basics: entity, legal, finance, tooling. Get those out of the way so they never become a bottleneck. Then you&#8217;re straight into the tightrope.</p><p style="text-align: justify;">Product and GTM are a symbiosis. You can&#8217;t build something nobody wants, and you can&#8217;t sell something that doesn&#8217;t exist yet. They need equal attention. The problem is you can&#8217;t give them equal attention at the same time. Some weeks you&#8217;re deep in product, building, testing, iterating. Other weeks you&#8217;re all GTM, talking to people, running experiments, figuring out your channels. The skill isn&#8217;t doing both. The skill is knowing when to lean into one and pull back on the other without letting either one fall over.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!bmLu!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!bmLu!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png 424w, https://substackcdn.com/image/fetch/$s_!bmLu!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png 848w, https://substackcdn.com/image/fetch/$s_!bmLu!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png 1272w, https://substackcdn.com/image/fetch/$s_!bmLu!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png 1456w" sizes="auto, 100vw"><img src="https://substackcdn.com/image/fetch/$s_!bmLu!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png" width="894" height="544" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/b6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:544,&quot;width&quot;:894,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:886680,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://thirdv3ctor.substack.com/i/193482178?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!bmLu!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png 424w, https://substackcdn.com/image/fetch/$s_!bmLu!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png 848w, https://substackcdn.com/image/fetch/$s_!bmLu!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png 1272w, https://substackcdn.com/image/fetch/$s_!bmLu!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb6d0838f-815c-4b30-8efc-a6dbfcc44f60_894x544.png 1456w" sizes="auto, 100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p style="text-align: justify;"></p><p style="text-align: justify;">In Dutch we call this <em>koorddansen</em>. Tightrope walking. And it&#8217;s the thing we see founders get wrong more than almost anything else. Either they build for months without talking to anyone, or they sell something they haven&#8217;t figured out how to deliver yet.</p><p style="text-align: justify;">And the thing that ties all of it together is talking to prospects. That never stops. Every product decision, every GTM move comes back to what we heard in the last conversation. You can have the best frameworks in the world, but if you&#8217;re not talking to the people you&#8217;re building for, you&#8217;re guessing with extra steps.</p><p style="text-align: justify;">The cadences need to line up. Operations, product, GTM. How often you review and plan, how fast you ship, how you find and close customers. When they&#8217;re out of sync you feel it immediately. You&#8217;re building something nobody asked for, or selling something that&#8217;s three sprints behind what you promised.</p><h2 style="text-align: justify;"><strong>What&#8217;s coming</strong></h2><p style="text-align: justify;">We&#8217;ve organized everything into four workstreams &#8212; GTM, Product, Finance, and Operations &#8212; all published under <em>The Blueishprint</em>.</p><p style="text-align: justify;">GTM and Product start now. Finance and Operations will come when they become relevant. Every article comes with templates, prompts, and the actual tools we use. This isn&#8217;t a locked roadmap. If you tell us you want something else first, we&#8217;ll adjust.</p><p style="text-align: justify;">Next week we kick off with getting your GTM strategy on paper. We&#8217;ll walk you through writing it, stress-testing it, and building the pipeline to actually execute it. From there the series goes into target lists, lead enrichment, CRM setup, sales sequences, and automated outreach.</p><p style="text-align: justify;">The Product series starts with how we capture user feedback and turn it into shipped code, with AI doing most of the grunt work. We&#8217;ll share the full setup, from feedback intake to Linear tickets to deployment.</p><p style="text-align: justify;">Beyond that: funding (do you actually need it), cap tables, hiring when AI has changed the game, and the personal stuff like how to actually take a holiday when you&#8217;re building a company.</p><p style="text-align: justify;">The one we&#8217;re most looking forward to writing: &#8220;I Got Acquired in a Stock Deal. Here&#8217;s What Nobody Told Me.&#8221; Stock deal vs. cash deal, tax traps, earnout games, and everything that blindsided us. Almost nobody writes about this from the founder&#8217;s side.</p><h2 style="text-align: justify;"><strong>How we write (and why process beats prompts)</strong></h2><p style="text-align: justify;">Full transparency: one of us is dyslexic. For most of his career, writing was the last thing he&#8217;d volunteer for. AI changed that. We use AI to help us write, to turn messy thinking into clear sentences, to catch errors our brains skip over.</p><p style="text-align: justify;">But the thing everyone gets wrong about AI right now? There&#8217;s an endless flood of articles about clever prompts and AI skills and &#8220;10x your productivity with this one trick.&#8221; <em>Open deur</em>, as we say in Dutch: garbage in, garbage out. A prompt is only as good as the thinking behind it. If you feed vague input into any tool, you get vague output.</p><p style="text-align: justify;">AI helps us say it clearly. The thinking is ours. We mention this because we think more founders should talk about it. Using AI to write isn&#8217;t cheating. It&#8217;s a tool, like everything else. And it&#8217;s one of the reasons <em>The</em> <em>Blueishprint</em> exists at all.</p><h2 style="text-align: justify;"><strong>The community</strong></h2><p style="text-align: justify;"><em>The Blueishprint</em> isn&#8217;t a newsletter you read and forget about. We want to build something where founders and operators raise real questions and work through them together. Someone struggling with their pricing gets input from ten people who&#8217;ve been through it. The articles are just the start.</p><p style="text-align: justify;">We have an open WhatsApp group you can join right now. It&#8217;s live, it&#8217;s messy, and people ask real questions in real time. No gatekeeping, no &#8220;premium tier.&#8221; Just people building things and helping each other figure it out.</p><p style="text-align: justify;"><a href="https://chat.whatsapp.com/CsnUENcdfvkFzp7tNxB4Ya?">Join here</a>.</p><h2 style="text-align: justify;"><strong>Your turn</strong></h2><p style="text-align: justify;">Here&#8217;s our question for you: what&#8217;s the one thing you&#8217;re stuck on right now in your business? The thing you keep circling back to but haven&#8217;t cracked yet. Drop it in the WhatsApp group, reply to this email, leave a comment. We&#8217;ll either write about it, or someone in the community already has an answer.</p><p style="text-align: justify;">And if you know someone who&#8217;s in the middle of building something right now, send this their way. The more people in the conversation, the better the answers get.</p><p style="text-align: justify;">See you next Tuesday.</p><p style="text-align: justify;">&#8212; Friso &amp; Tim</p><p style="text-align: justify;"><em>Images: Tristar Pictures, Ivan S (Pexels)</em></p>]]></content:encoded></item><item><title><![CDATA[Third Vector Observer | Edition 2]]></title><description><![CDATA[Claude Cowork Spooks Investors, Dead Companies Walking, the High-Growth Handbook]]></description><link>https://blueishprint.thirdvector.io/p/third-vector-observer-edition-2</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/third-vector-observer-edition-2</guid><dc:creator><![CDATA[Third Vector]]></dc:creator><pubDate>Fri, 06 Feb 2026 10:47:40 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/434fad2a-cdb9-4f8d-84c3-f15e7ee7bbb0_784x1168.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p></p><p class="button-wrapper" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe now&quot;,&quot;action&quot;:null,&quot;class&quot;:null}" data-component-name="ButtonCreateButton"><a class="button primary" href="https://blueishprint.thirdvector.io/subscribe?"><span>Subscribe now</span></a></p><p><strong>Welcome to the Third Vector Observer</strong>, an easy-to-digest newsletter featuring 3&#8211;5 high-quality, insightful pieces on venture operations, agentic AI, finance, and Europe's tech ecosystem. Delivered to your inbox roughly twice a month.</p><h2>Things We Observed&#8230;</h2><h3>1. Claude Cowork Triggers $285B Selloff Across Software, Financial Services, and Asset Management Stocks</h3><p>Anthropic&#8217;s release of <strong>Claude Cowork</strong> spooked markets in January, but it was last week&#8217;s launch of 11 open-source plugins that turned a tremor into an earthquake. By February 3rd, a massive <strong>$285B selloff</strong> hit software, financial services, and asset management stocks globally.</p><p><a href="https://www.siliconrepublic.com/business/anthropics-new-cowork-plug-ins-prompt-sell-off-in-software-shares">Why the panic?</a></p><p>The market has reached a tipping point. Investors are no longer viewing AI as a "feature" for existing SaaS; they are viewing it as a <strong>replacement for the application layer</strong>. When Anthropic&#8212;an infrastructure provider&#8212;released ready-made vertical solutions for legal, finance, and sales, the platform officially became the competitor.</p><p></p><h4><strong>What makes Cowork different: agentic autonomy</strong> </h4><p>Claude Cowork doesn't just chat; it executes multi-step workflows directly on a user&#8217;s machine.</p><p><a href="https://support.claude.com/en/articles/13345190-getting-started-with-cowork">Cowork includes a library of plugins</a> for common knowledge work functions:</p><ul><li><p><strong>Productivity</strong> &#8212; Manage tasks, calendars, and daily workflows</p></li><li><p><strong>Enterprise search</strong> &#8212; Find information across your company&#8217;s tools and docs</p></li><li><p><strong>Sales</strong> &#8212; Research prospects, prep deals, and follow your sales process</p></li><li><p><strong>Finance</strong> &#8212; Analyze financials, build models, and track key metrics</p></li><li><p><strong>Data</strong> &#8212; Query, visualize, and interpret datasets</p></li><li><p><strong>Legal</strong> &#8212; Review documents, flag risks, and track compliance</p></li><li><p><strong>Marketing</strong> &#8212; Draft content, plan campaigns, and manage launches</p></li><li><p><strong>Customer support</strong> &#8212; Triage issues, draft responses, and surface solutions</p></li><li><p><strong>Product management</strong> &#8212; Write specs, prioritize roadmaps, and track progress</p></li><li><p><strong>Biology research</strong> &#8212; Search literature, analyze results, and plan experiments</p></li></ul><p></p><h4>&#8220;Your Margin is My Opportunity&#8221;</h4><p>The damage was swift and targeted at companies whose business models rely on &#8220;seat-based&#8221; knowledge work:</p><ul><li><p><strong>Legal Tech:</strong> Thomson Reuters (-16%), RELX/LexisNexis (-14%), and Wolters Kluwer (-13%) saw massive wipes.</p></li><li><p><strong>Enterprise SaaS:</strong> Heavyweights like Salesforce and ServiceNow fell 7%, as the threat of &#8220;seat compression&#8221; became real.</p></li><li><p><strong>Outsourcing:</strong> Indian IT giants (Infosys, Wipro) slid 5-6% as investors feared agentic AI would cannibalize FTE-based contracts.</p></li></ul><p></p><p>Software ate the world, now AI is eating software. As Jeff Bezos famously said, <em>&#8220;Your margin is my opportunity.&#8221;</em> In 2026, AI isn't just eating software; it's eating those fat, juicy <strong>SaaS margins</strong>. Yum.</p><p></p><h3>2. Dead Companies Walking - Lessons from a Short-Seller Applied to SaaS</h3><p>In <a href="https://tomtunguz.com/dead-companies-walking/">this sobering piece</a>, <strong>Tomasz Tunguz</strong> applies the &#8220;six failure modes&#8221; of legendary short-seller <strong>Scott Fearon</strong> (<em>Dead Companies Walking</em>) to the current state of the software industry. Tunguz&#8217;s thesis is clear: the &#8220;hairpin turn&#8221; from SaaS to AI is amplifying every classic mistake that causes companies to zero out.</p><p><strong>Fearon&#8217;s Six Failure Modes (Applied to the AI Era):</strong></p><ol><li><p><strong>Learning only from the recent past:</strong> SaaS multiples have been flat for three years. Relying on 2021-era growth playbooks is the fastest way to run out of runway.</p></li><li><p><strong>Relying too heavily on a formula:</strong> The &#8220;efficient growth&#8221; formulas of the last decade (PLG vs. Enterprise) are breaking. In an AI-native world, sales motions and quotas must be rebuilt from scratch.</p></li><li><p><strong>Misreading or alienating customers:</strong> Customers don&#8217;t want &#8220;SaaS with an AI wrapper&#8221;; they want AI-native solutions that solve pain points in fundamentally new ways. Selling the old application leads to instant churn.</p></li><li><p><strong>Falling victim to a mania:</strong> Shipping half-baked AI features to appease a roadmap isn&#8217;t the same as delivering value. As Tunguz notes, 2026 is the year where &#8220;long-running agents&#8221; will separate the wheat from the chaff.</p></li><li><p><strong>Failing to adapt to tectonic shifts:</strong> The shift from seat-based pricing to outcome-based or compute-based models is a structural earthquake.</p></li><li><p><strong>Being physically or emotionally removed:</strong> If leadership isn&#8217;t using tools like Claude or GPT daily, they aren&#8217;t just behind&#8212;they are &#8220;emotionally removed&#8221; from the reality of the technology.</p></li></ol><p>Link: https://tomtunguz.com/dead-companies-walking/</p><p></p><h3>3. The High-Growth Handbook: Molly Graham on Lenny&#8217;s Podcast</h3><p>In this episode of Lenny&#8217;s Podcast, Molly Graham (Glue Club, ex-Facebook/Google) shared a masterclass on leading through the chaos of hyper-growth. For anyone building or advising ventures, her frameworks and mental models help moving from &#8220;scrappy startup&#8221; to &#8220;scaled machine.&#8221; See selected highlights below the Spotify link.</p><iframe class="spotify-wrap podcast" data-attrs="{&quot;image&quot;:&quot;https://i.scdn.co/image/ab6765630000ba8ad50137f690dac4b9e789da88&quot;,&quot;title&quot;:&quot;The high-growth handbook: Molly Graham&#8217;s frameworks for leading through chaos, change, and scale&quot;,&quot;subtitle&quot;:&quot;Lenny Rachitsky&quot;,&quot;description&quot;:&quot;Episode&quot;,&quot;url&quot;:&quot;https://open.spotify.com/episode/6i0638I1MHCa7WAAgkhrUu&quot;,&quot;belowTheFold&quot;:true,&quot;noScroll&quot;:false,&quot;isVideo&quot;:false}" src="https://open.spotify.com/embed/episode/6i0638I1MHCa7WAAgkhrUu" frameborder="0" gesture="media" allowfullscreen="true" allow="encrypted-media" loading="lazy" data-component-name="Spotify2ToDOM"></iframe><h4><strong>The Waterline Model</strong> </h4><p>Molly shared a useful framework, the <strong>&#8220;Waterline Model&#8221;,</strong> for diagnosing team dysfunction. She argues that when a team is struggling, leaders often default to &#8220;scuba diving&#8221;&#8212;going deep into interpersonal conflicts or individual performance issues. Instead, she suggests you should <strong>&#8220;snorkel&#8221;</strong> first.</p><ul><li><p><strong>Above the Waterline:</strong> These are structural issues&#8212;unclear roles, lack of alignment on goals, or poor processes.</p></li><li><p><strong>Below the Waterline:</strong> These are the messy, interpersonal &#8220;people&#8221; problems.</p></li></ul><p><strong>The Insight:</strong> 80% of what we perceive as &#8220;personality clashes&#8221; or &#8220;bad hires&#8221; are actually just structural failures above the waterline. Before you blame the person, fix the role clarity and the goal-setting. It&#8217;s a powerful reminder for venture operators: <strong>Structure dictates behaviour.</strong></p><p></p><h4><strong>Stop Over-Investing in Low Performance</strong></h4><p>Molly&#8217;s advice centers on the <strong>opportunity cost</strong> of a manager&#8217;s time. She argues that the common instinct&#8212;spending 80% of your energy on the bottom 10%&#8212;is a strategic error.</p><blockquote><p>&#8220;The mistake most managers make is they spend all their time on their lowest performers... it&#8217;s actually a disservice to the rest of the team. You have to get comfortable with the fact that if it&#8217;s not working, the kindest thing you can do is move them out quickly. Then, take all that energy and pour it into your &#8216;high-potentials&#8217;&#8212;the people who are actually going to build the future of the company.&#8221;</p></blockquote><p></p><p><strong>Other Key Takeaways:</strong></p><ul><li><p><strong>&#8220;Give Away Your Legos&#8221;:</strong> The classic framework for scaling yourself. If you aren&#8217;t actively handing off the parts of your job you love, you&#8217;re becoming the bottleneck.</p></li><li><p><strong>The &#8220;J-Curve&#8221; Career:</strong> Why the most successful paths often look like a series of terrifying leaps into the unknown rather than a steady climb up a corporate ladder.</p></li><li><p><strong>Strategy Should Hurt:</strong> If your strategy doesn&#8217;t involve making painful trade-offs (saying &#8220;no&#8221; to good ideas), it&#8217;s not a strategy; it&#8217;s a wishlist.</p></li></ul><div><hr></div><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Third Vector! Subscribe for free to receive new posts and follow our work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item><item><title><![CDATA[Tim’s Favourite Books of 2025 ]]></title><description><![CDATA[Warning: Eclectic Mix]]></description><link>https://blueishprint.thirdvector.io/p/tims-favourite-books-of-2025</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/tims-favourite-books-of-2025</guid><dc:creator><![CDATA[Tim de Rooij]]></dc:creator><pubDate>Tue, 03 Feb 2026 18:57:06 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!l3hP!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc6c9beb9-d47d-4d1f-be40-e999d56ea1d2_1024x1024.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>My 2025 reading list was more eclectic than in years past&#8212;a blend of biography, fantasy, science fiction, and business strategy.</p><p>These were my five favourite reads from the last year:</p><h2><strong>Elon Musk </strong>(Biography) by Walter Isaacson</h2><p>I think it is fair to say that Elon Musk evokes strong emotional responses. There are some who admire his brilliance in entrepreneurship, building industry-disrupting companies ranging from Tesla to SpaceX and xAI. There are others who despise him for his alt-right rhetoric and confrontational approach to, yeah, pretty much everything. He is the quintessential subject for a biography, and Isaacson has delivered a masterwork.</p><h2><strong>The Hobbit</strong> by J.R.R. Tolkien</h2><p>Though I&#8217;ve seen the films, I had never explored Tolkien&#8217;s original prose. I started with <em>The Hobbit</em>, as one does, and found the world-building as immersive as the legends suggest.</p><h2><strong>Ruud Lubbers: Een slag anders</strong> by Johan van Merri&#235;nboer</h2><p>Admittedly, you need to be a bit of a geek on Dutch Politics to fully appreciate this one. Lubbers served as Prime Minister of the Netherlands for a 12-year period, from 1982 to 1994, navigating the country through a period of economic recession, the missile crisis, and European integration. He played a key role in negotiations with Reagan and Gorbachev to put a halt to nuclear proliferation. The book is both a biography of a natural leader with an unparalleled work ethic and a historical account of the final decade of the Cold War.</p><h2><strong>How Big Things Get Done</strong> by Bent Flyvbjerg</h2><p>A mind-blowing 92% of all projects fail to be delivered within budget and on time. When measuring success strictly as on-time, on-budget delivery of promised benefits, only 0.5% (or 1 in 200 projects) succeed&#8212;leaving 99.5% as failures in all three metrics. Bent Flyvbjerg is a Danish economic geographer and leading expert on megaprojects. In How Big Things Get Done he shares key lessons from his research on thousands of failed and successful megaprojects. Flyvbjerg explores cases ranging from the Sydney Opera House (an epic failure) to the Empire State Building (a resounding success, finished 17% under budget). A fun and insightful read for everyone whose job entails leading and executing big or small projects.</p><h2><strong>Project Hail Mary</strong> by Andy Weir</h2><p>From the author of <em>The Martian</em>, this is a high-stakes survival story. It follows Ryland Grace, a teacher-turned-astronaut tasked with saving Earth from a solar-dimming threat. It&#8217;s a masterclass in creative problem-solving. And it turns out humans aren&#8217;t the only species fighting for survival.</p>]]></content:encoded></item><item><title><![CDATA[Third Vector Observer | Edition 1]]></title><description><![CDATA[AI Adoption, Virtual Advisors, GTM in the AI Era, Typeless, Agentic AI Foundations]]></description><link>https://blueishprint.thirdvector.io/p/third-vector-observer-edition-1</link><guid isPermaLink="false">https://blueishprint.thirdvector.io/p/third-vector-observer-edition-1</guid><dc:creator><![CDATA[Third Vector]]></dc:creator><pubDate>Thu, 08 Jan 2026 23:03:19 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/34dcdd95-c06f-445c-883c-ac1d9f633287_1000x1000.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p></p><p class="button-wrapper" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe now&quot;,&quot;action&quot;:null,&quot;class&quot;:null}" data-component-name="ButtonCreateButton"><a class="button primary" href="https://blueishprint.thirdvector.io/subscribe?"><span>Subscribe now</span></a></p><p><strong>Welcome to the first edition of Third Vector Observer</strong>, an easy-to-digest newsletter featuring 5&#8211;10 high-quality, insightful pieces on venture building, AI, and news from Europe's tech ecosystem. Delivered to your inbox roughly twice a month.</p><h2>Things We Observed&#8230;</h2><h3>1. The Economist: AI Adoption </h3><p>A reality check on the state of <strong>AI adoption</strong>&#8212;and a reminder that not everyone is leaning in just yet.</p><p>&#8220;Adoption has fallen sharply at the largest businesses, those employing over 250 people. Three years into the generative-AI wave, demand for the technology looks surprisingly flimsy.&#8221;</p><p>And:</p><p>&#8220;From today until 2030 big tech firms will spend $5trn on infrastructure to supply AI services. To make those investments worthwhile, they need on the order of $650bn a year in AI revenues, according to JP Morgan Chase, a bank, up from about $50bn a year today.&#8221;</p><p><a href="https://www.economist.com/finance-and-economics/2025/11/26/investors-expect-ai-use-to-soar-thats-not-happening">The Economist - Investors expect AI use to soar, that's not happening</a> (possibly behind a paywall)</p><h3>2. Fred Wilson: Making AI Virtual Advisors</h3><p>I&#8217;ve been reading Fred Wilson&#8217;s blog, <em><strong><a href="https://avc.xyz/">AVC</a></strong></em>, for over 15 years. As a founding partner of Union Square Ventures, Fred pioneered blogging in the VC industry, sharing insights on tech trends, the state of VC, and practical advice for entrepreneurs. He maintains a massive following, even after moving his blog &#8220;onchain&#8221; (to the blockchain).</p><p>When I started working with startups ~10 years ago, I benefited from his frameworks on cap table management, startup valuation, and emerging tech like machine learning. </p><p>In this post, he discusses creating AI virtual advisors. It resonates because I&#8217;ve created a virtual CFO, CMO, and CTO using Google&#8217;s Gemini (Gems) for Third Vector. Interacting with these advisors <strong>every day</strong> is a true productivity multiplier.</p><p><a href="https://avc.xyz/making-advisors">AVC - Making Advisors</a></p><h3>3. GTM in the AI Era </h3><p>Lenny&#8217;s Podcast is a pure treasure trove. Lenny is the chillest, most humble host in the space. This episode with Vercel&#8217;s COO Jeanne DeWitt Grosser (formerly of Stripe and Google) on what GTM looks like in 2026 is packed with insights.</p><p>To pick just one, in three simple steps Jeanne&#8217;s team built a <strong>&#8220;Lostbot&#8221;</strong> to learn from lost deals:</p><p>Step 1: Pull a list of lost deals from the previous quarter by size.<br>Step 2: Gather all sales transcripts (Vercel uses Gong), Slack convos, and email threads.<br>Step3: Use an AI agent to diagnose each deal and extract learnings.</p><p>They then used those lessons to create a <strong>&#8220;Dealbot&#8221;</strong> that analyzes live deals to flag risks&#8212;like, &#8220;Hey, you haven&#8217;t talked to the economic buyer yet.&#8221; Essentially, these bots help Vercel find bugs in their go-to-market process.</p><iframe class="spotify-wrap podcast" data-attrs="{&quot;image&quot;:&quot;https://i.scdn.co/image/ab6765630000ba8a1ee38de7320448adc9a107b6&quot;,&quot;title&quot;:&quot;What world-class GTM looks like in 2026 | Jeanne DeWitt Grosser (Vercel, Stripe, Google)&quot;,&quot;subtitle&quot;:&quot;Lenny Rachitsky&quot;,&quot;description&quot;:&quot;Episode&quot;,&quot;url&quot;:&quot;https://open.spotify.com/episode/33oMOus6SMZvaN6CFKpXRo&quot;,&quot;belowTheFold&quot;:true,&quot;noScroll&quot;:false,&quot;isVideo&quot;:false}" src="https://open.spotify.com/embed/episode/33oMOus6SMZvaN6CFKpXRo" frameborder="0" gesture="media" allowfullscreen="true" allow="encrypted-media" loading="lazy" data-component-name="Spotify2ToDOM"></iframe><h3>4. AI dictation with Typeless</h3><p>AI-native voice-typing got really, really good. It still feels a bit weird to talk to your computer, but it&#8217;s incredibly fast (4x faster than typing) and the transcription quality is stunning. I&#8217;ve been using <a href="https://www.typeless.com/">Typeless</a> for note-taking in Obsidian and for drafting emails; I love it. They offer a free tier that is well <a href="https://www.typeless.com/">worth checking out</a>.</p><h3>5. Gaming and AI Research Lab with Dutch roots Raises $133.7M in Seed</h3><p>It turns out video game recordings are a gold mine for training AI on scenarios that are expensive to simulate&#8212;like helicopter crashes or plane accidents. Pim de Witte, the Dutch founder of <a href="https://medal.tv/">Medal.tv</a> (a platform that beats YouTube in annual video game clip uploads), decided to spin off a dedicated research lab for foundation models trained on gaming footage.</p><p>The lab, called <strong><a href="https://www.generalintuition.com/">General Intuition</a></strong>, raised <a href="https://techcrunch.com/2025/10/16/general-intuition-lands-134m-seed-to-teach-agents-spatial-reasoning-using-video-game-clips/">$133.7M in seed funding</a> to build models for use cases requiring spatial and temporal reasoning, such as autonomous vehicles and robotics. While now headquartered in NYC, General Intuition&#8217;s parent company is based in Naarden, Netherlands.</p><h3>6. Protocols and Agentic AI Foundation </h3><p>This article from <em>The Economist</em> (Dec 13, 2025) does an excellent job explaining how different AI protocols, specifically A2A and MCP, work together to create the next version of the web. </p><p><a href="https://www.economist.com/interactive/science-and-technology/2025/12/10/the-next-version-of-the-web-will-be-built-for-machines-not-humans">The Economist - The next version of the web will be built for machines not humans</a></p><p></p><div><hr></div><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blueishprint.thirdvector.io/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thanks for reading Third Vector! Subscribe for free to receive new posts and follow our work.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><p></p>]]></content:encoded></item></channel></rss>