Somewhere around my third new venture in as many months, I caught myself doing something that should have bothered me sooner.
I was rebuilding the same thinking from scratch. Again.
Evaluate the idea. Sketch a business model. Map out go-to-market. Work through pricing. Each time, I’d dig through old notes, half-remember a framework from something I’d watched two years ago, piece together something I’d already built before, just less precisely than the first time. The output was fine. The process was broken.
The frustrating part wasn’t the effort. It was the waste. I’d done this thinking before. I’d refined it across multiple companies. Somewhere in 18 years of building, I’d developed real conviction about how to validate an idea quickly, how to structure a cap table for a seed round, how to think about pricing before you have enough data to be confident. That knowledge existed. It just lived in my head, scattered across notes, and got partially reconstructed every single time I needed it.
That’s not a system. That’s a liability.
The real problem AI didn’t solve
When the current generation of AI tools arrived, I watched a lot of founders breathe a sigh of relief. Building got faster. Code got cheaper. Idea-to-prototype timelines collapsed. The tools genuinely worked.
But I kept noticing something. The founders I was talking to, across Dubai, across the broader ecosystem I operate in, were moving faster without necessarily thinking better. They were using AI to accelerate execution while still doing the underlying strategic thinking the old way: from instinct, from half-remembered advice, from whatever framework they’d last read about.
AI made building accessible. It didn’t make thinking systematic.
The bottleneck shifted. It didn’t disappear.
What I built instead
I built Startup OS. Twelve structured AI skills, each one a curated knowledge pack that turns Claude into a domain expert for a specific phase of building. Not a chatbot that gives generic advice. Not a course. Not a PDF that sits in your downloads folder.
A system that runs frameworks on your inputs.
The distinction matters. When you load the idea validation skill and tell it about your startup, it doesn’t explain the Mom Test to you. It designs interview scripts for your specific idea, with your specific target customer, then scores the results against a GO/PIVOT/KILL framework. When you load the finance skill and give it your numbers, it doesn’t explain what burn rate means. It calculates yours and tells you when you run out of money.
The twelve skills cover the full arc of building: idea validation, business model design, fundraising, go-to-market, product management, founder-led sales, marketing and brand, growth and analytics, operations, finance, customer success, and legal and compliance, including entity structure and cap table mechanics adapted for UAE/MENA and crypto contexts.
Every skill draws from methodologies I’ve actually used. Rob Fitzpatrick’s Mom Test. Strategyzer’s business model patterns. How VCs actually evaluate deals, which is cash-on-cash math, not narrative. Lenny Rachitsky’s research on how 40+ companies found their first customers. MEDDIC and Challenger for sales. RICE prioritization for product. AARRR benchmarks for growth.
These aren’t frameworks I’ve bookmarked. They’re frameworks I’ve run.
Why this is different from just reading about frameworks
There are two things that make a system useful rather than decorative.
First: it has to be interactive, not informational. The skills don’t present knowledge. They apply it. There’s a meaningful difference between knowing that burn rate matters and having something that calculates yours in real time based on the actual numbers in front of you. One gives you a concept. The other gives you a decision.
Second: it has to be contextual. When you load multiple skills into a Claude Project alongside your own documents, your pitch deck, your financial model, your customer research, the AI connects them. Your business model informs your go-to-market. Your go-to-market shapes your financial projections. Your projections anchor your fundraising strategy. The frameworks stop being isolated chapters and start functioning as an integrated operating system.
That’s the part that took time to design properly. Individual skills are useful. The system is what compounds.
Who this is for
I built this for myself first. The founder who has enough instinct to move but wants frameworks running alongside that instinct, checking blind spots, surfacing questions worth asking, turning pattern recognition into repeatable process.
If you’re building your first venture, the system gives you operator-grade frameworks you’d otherwise spend years developing through trial and error. If you’ve built before, it gives you the structured version of what you already know, ready to run instead of ready to reconstruct.
The full Startup OS is available as a download: twelve SKILL.md files, setup instructions for Claude Projects and Cowork, and a reference guide for how to invoke specific skills depending on where you are in the build. The files are plain text, not locked software. If a framework stops serving you, you swap it out. The system is designed to grow with your thinking, not stay fixed around mine.
Eighteen years of building taught me that instinct and frameworks compound faster together than either does alone. This is that combination, made into something you can actually run.
