The AI Skill That Rewired How I Think Like a Founder
gstack's office-hours skill doesn't help you build faster. It helps you build the right thing by forcing the questions most founders are afraid to ask.
The first time I ran /office-hours, I expected it to help me move faster.
It did the opposite. It made me stop.
Not in a frustrating way — in the way a good mentor does when you walk in excited about an idea and they say: okay, slow down. Tell me who specifically has this problem.
That pause is worth more than most code I've ever written.
The Skill That Asks the Uncomfortable Questions
gstack's office-hours skill is part of Garry Tan's gstack toolkit — a collection of 35 specialist skills that turn Claude Code into a full virtual engineering team. But office-hours isn't a technical skill. It's a thinking skill.
It has two modes.
Startup Mode is for founders, product managers, and anyone building something that needs to survive contact with paying users. It runs you through six questions that are deceptively simple on the surface and genuinely hard to answer with honesty:
- What is the exact problem? Who specifically has it?
- How are they solving it today — and what does that cost them in time, money, or pain?
- Who is the most desperate person in this market?
- What is the smallest version of this product that proves the core insight?
- Have you actually watched someone struggle with this problem in real life?
- Will this be more important or less important in three years?
Question five is the one that gets most people.
It is very easy to have opinions about user problems. It is much harder to have observations. The difference between the two is the gap between products that get traction and products that don't.
Builder Mode is for side projects, open-source experiments, and anything driven by curiosity rather than business logic. It doesn't pressure you with monetization questions — it helps you find the most electric version of your idea. What makes it genuinely interesting? What's the thing that would make someone's jaw drop?
This matters because not every good project starts with a problem. Some start with a question or a hunch. Builder Mode respects that while still pushing you to find a sharp angle.
Why Most Founders Skip This Step
There's a particular kind of excitement that comes with a new idea. It's one of the best feelings in building — the rush of possibility, the mental movie of what this could become.
The danger is that excitement and evidence feel identical from the inside.
When you're in that state, your brain is extremely good at generating supporting arguments. The market is huge. The pain point is obvious. Nobody has solved it well. You've heard five people complain about this exact thing.
These might all be true. But they're not the same as knowing who your first twenty users are, why they'd switch away from their current solution today, and what they'd lose if you disappeared tomorrow.
office-hours is designed to interrupt that pattern. It creates just enough friction to separate genuine insight from manufactured confidence.
Garry Tan runs YC now, and that perspective shows up in the design. The questions aren't novel — they're the same ones every YC partner asks in every office hours session. The skill is essentially a YC partner in a prompt: patient, rigorous, impossible to bullshit.
The Design Doc at the End
What I didn't expect was how useful the output is.
After the conversation, office-hours generates a structured design document capturing your answers — the real user, the current workaround, the pain cost, the minimum viable form, and the directional bet.
This sounds like a small thing. It isn't.
The document forces your thinking from a loop in your head into a written claim you have to defend. That shift matters enormously. Written claims are testable. You can share them, pressure-test them, return to them six weeks later and ask: did I actually validate this?
Most founders never write this document. They carry their thesis loosely in their head, update it unconsciously, and wonder why the team is misaligned or why the roadmap keeps shifting. The design doc is the foundation. office-hours builds it with you in twenty minutes.
A New Pre-Build Ritual
I now use office-hours before starting any significant new project — not just business ideas, but features, content series, and technical tools.
The questions are always the same:
- Is this problem real, or do I just find it interesting?
- Who is the person who needs this most urgently?
- What are they doing right now, and what is that costing them?
- Can I build something testable in a week?
- Is this more important in three years, or is this a moment in time?
If I can't answer these clearly, I shouldn't be writing code. If I can answer them, everything I build after is sharper.
That's the function office-hours serves. Not to help you build faster. To help you build the thing that deserves to be built.
The fastest progress isn't more code. It's being certain you're building the right thing before you start.
AI Skill Daily 002. Part of the gstack series — 35 specialist skills from garrytan/gstack.