Tools: We used Patrick to make Patrick. No this is not another LLM story.

What Patrick actually is ## The problem nobody's talking about ## How it actually works ## The moment it got recursive ## Patrick told us what Patrick needed. ## Why we're releasing it ## What this means for you Well, that’s a partial lie. But let me set the scene. My colleagues and I are working on our medtech solution. There are only a handful of us, and every iteration of the software, every client interview and every new partnership opens up tens of possible directions. Should we chase the hospital pilot or the telehealth integration? Do we build the mobile UX first, or lock down compliance? That clinic in Saskatchewan wants something slightly different from the research hospital in Winnipeg - do we fork the roadmap or find the overlap? We were keeping track of all of this the way most startups do: scattered notes, shared docs, the occasional spreadsheet that someone updates heroically for a week and then abandons. It worked until it didn't. One day we sat down for a strategy meeting and realized we had forty-plus organizations in our orbit, five product lines at various stages, over a hundred open tasks, and no coherent way to see how any of it connected. So we built Patrick. Patrick is not a dashboard. It's not a CRM. It's not a project management tool, though it can behave like all three when you need it to. At its core, Patrick is a structured summary, and a knowledge graph. Patrick doesn’t summarize things for the purpose of getting the word count down or making it faster to read, Patrick summarizes documents with purpose; “how would this task’s development contribute to this initiative?” or “how does this organization’s need justify this development task, and are they aligned with the priorities set in the shareholders meeting?” are just examples of how Patrick looks at things. Concepts connect to each other in meaningful ways, and the connections are the point. When we ask “what happens if we delay this feature?” Patrick doesn’t just show us a Gantt chart turning red. It traces the impact upstream and downstream, which customer needs go unmet, which prospects are affected, which tasks become orphaned. When we ask “what should we build next?” it doesn’t just sort by priority, it weighs value against effort against risk and tells us where the quick wins are hiding and where the strategic bets live. The thing that makes it different from a spreadsheet, a Notion board or an Obsidian vault is that Patrick understands relationships. Here's the thing everyone gets wrong about AI tools. The conversation is always about the model. Which one is smarter, which one is faster, which one hallucinates less. But the model isn't the bottleneck anymore. You are. Or more precisely, what you tell it is. Think about the last time you asked an LLM to help you make a decision about your business. You probably spent ten minutes writing a prompt that tried to capture the full picture - who your clients are, what you're building, which deals are in play, what's blocking what. You gave it a slice of the truth, and it gave you a confident answer based on that slice. Maybe it was useful. Maybe it missed the thing you forgot to mention. Now think about the best executive you've ever worked with. When they sit down to think through a problem, they're not working from a prompt they typed in five minutes ago. They're working from a mental model of the entire organization - every relationship, every dependency, every half-finished initiative, every promise made to a client six months ago. That context is what makes their judgment good. Patrick is that context, externalized and structured so an AI can use it the way a great executive uses institutional memory. The insight that led to Patrick isn't technical. It's this: the people who get the most out of AI aren't the ones with the best prompts. They're the ones who've figured out how to feed the AI a true and complete picture of their situation. We studied what those people do. The executives and operators who consistently get strategic value out of LLMs, not just help writing emails. What we found is that they all do some version of the same thing: they maintain structured information about their business and inject it into their conversations with AI. Some do it with elaborate Notion setups. Some do it with custom GPTs stuffed with documents. Most do it badly, or inconsistently, because maintaining that structure by hand is a second job nobody signed up for. So we took what works and turned it into a system. Patrick gathers information about your organization - your prospects, your products, your team's capacity, your strategic priorities, the relationships between all of it - and structures it into a graph. Then, when you ask a question, the AI doesn't get a cold prompt. It gets the full organizational picture, tuned to the specific question you're asking. "Should we pursue this partnership?" isn't answered in a vacuum. It's answered in the context of what you're already building, who else needs the same capability, what it would cost, and whether it aligns with the direction you committed to last quarter. A few months in, Patrick had become the nervous system of our company. Every meeting started with "what does Patrick say?" Every new lead got entered as an organization with needs linked to features. Every week we'd run a portfolio health check and a value analysis. Then came the question: what should we build next for Patrick itself? We had ideas. Lots of them. A chatbot interface so non-technical teammates could query it conversationally. Better reporting templates. An evaluation framework for scoring strategic initiatives. So we did what had become instinct. We opened Patrick, created features for each idea, linked them to the needs they'd serve, estimated the effort, and ran the analysis. It surfaced that the conversational interface would unlock the most value - not because it was the most technically impressive, but because it would let our CEO and business development lead query the system directly instead of asking me to run it. That single insight reframed our entire roadmap. We built Patrick for a medtech company of five people managing forty-plus relationships across four provinces. But the problem it solves isn't a medtech problem. It isn't even a startup problem. Every knowledge worker using AI today is working with the same handicap: the AI is only as good as what you tell it, and nobody has time to tell it everything. Every prompt is a lossy compression of your actual situation. The more complex your work, the more you leave out, and the worse the output gets. Patrick is soon to be available as a skill on OpenClaw. That means if you're already running an OpenClaw instance - your own AI assistant on your own machine - you can install Patrick and start building a structured picture of your organization that your AI can actually use. You don't need to be technical. The skill comes with pre-built prompts crafted from patterns we've seen work - the same approaches that successful operators use to get strategic value out of AI, packaged so you don't have to figure them out yourself. Tell Patrick about your business. Feed it your meeting notes, your client list, your product roadmap. It structures it, connects it, and makes it available to your AI so that every question you ask is answered with the full picture. The result isn't a better chatbot. It's a better-informed one. And the difference between those two things is the difference between an AI that writes nice paragraphs and one that actually helps you think. If you've ever wished your AI assistant actually understood your business - not in the vague, "I'll pretend I remember" way, but in the "I know your three biggest prospects, what each of them needs, and which of your features satisfies two of them at once" way - that's what Patrick does. We used it to build a medtech company. Then we used it to build itself. Now we want to see what you build with it. Patrick is soon to be available as an OpenClaw skill. Install it, teach it your business, and start asking better questions. https://patrickbot.io Templates let you quickly answer FAQs or store snippets for re-use. Are you sure you want to hide this comment? It will become hidden in your post, but will still be visible via the comment's permalink. Hide child comments as well For further actions, you may consider blocking this person and/or reporting abuse