🧠Everyone’s building with AI. But fewer are using it to think — and that’s where the real leverage is.
I’ve used AI for plenty of tactical tasks — rewriting headlines, summarizing research, even coding and debugging. But where it’s been most transformational for me is in thinking work: strategy, messaging, positioning, and decision-making.
The kind of work that usually requires a whiteboard, a sharp colleague, and an afternoon of undivided focus.
Instead, I find myself doing it alone, in a single browser tab.
That’s not because I’m especially clever. It’s because I’ve gotten hooked on prompts.
Prompting is the new thinking tool
The biggest mistake people make with AI is assuming it will think for them. It won’t.
But if you start out with the right scaffolding — clear context, layered constraints, iterative refinement — it becomes an amazing kind of externalized thought partner. Faster than a collaborator. More readily available than a mentor. Relentless in its willingness to generate, critique, and reframe.
This first step is what you could describe as mental architecture: the practice of structuring your inquiry to deepen your own judgment.
The smartest technologists (including both founders and investors) I know are using AI in this way — not just to automate or ideate, but to actively shape their thinking. And the ones who aren’t? They’re stuck — not advancing while waiting for clarity — while others are shipping.
How AI Changes the Speed of Startup Thinking
There’s something surreal about hitting enter on a startup idea and watching a GTM plan materialize seconds later. It’s thrilling and candidly a little unnerving too.
It’s the amazing feeling of sketching on an infinite whiteboard — but the whiteboard can talk back, critique my thinking, draft a strategy or research brief while I sleep, and then generate a bunch of code to implement an idea.
The output is frequently better than any brainstorming sessions I’ve ever had.
In the past, startup strategy felt like chiseling stone. Now it feels like clay —shapeable, collaborative, and fast.
The real shift isn’t the code or the copy. It’s the feedback loop. Strategy used to take months. Now, you can make forward progress in minutes.
Faster loops change how we think. This isn’t about replacing human judgment. It’s about compounding it.
When the loop from idea → refinement → validation → GTM shrinks, you can take more shots on goal — with less sunk cost and more creative freedom.
15 Prompt Patterns for Startup Builders
If you're curious where to begin, here are 15 tested prompt strategies drawn from experimentation, Reddit, Hacker News, and real founder workflows of founders in my portfolio. These aren’t parlor tricks — they’re mental scaffolding at warp speed — and if you’re skeptical, try a few and tell me if your viewpoint changes.
1. Ask for a ranked list of problems in a domain you know
"What are the top 10 unsolved problems for GTM teams at fast-scaling startups?"
This prompt narrows the scope of what you’re looking for and boosts the signal of your response. I find that more specificity almost always gives vastly better results.
2. Constrain by pricing & business model
"Give me enterprise SaaS ideas that could charge $10K/year and don’t require a sales team. Please provide ideas in these 3 domains (specify ones you’re interested in), give me a clear problem articulation, an estimate of the US TAM, suggested GTM approach, and willingness to pay."
Clear constraints lead to sharper ideas.
3. Let the LLM learn your preferences first
"Before you suggest ideas, ask me 10 questions about the kind of business I want to build."
This can dramatically improve output.
4. Cross-pollinate sectors
"What enterprise collaboration tools might work in a healthcare context?"
Every LLM I’ve tested is excellent at lateral thinking.
5. Go downstream from hot trends
"What new analytics capabilities will AI-native developer platforms need in 6–12 months?"
Using AI for mapping the support ecosystem saves me hours of analyst work.
6. Work backwards from your unfair advantage
"What startup ideas fit someone who’s ex-AWS, familiar with developer platforms, and has deep clinical workflow knowledge?"
Make your edge the starting point.
7. Prompt for positioning and GTM, not just ideas
"Here are 3 devtool ideas. What’s a differentiated GTM strategy for each?"
Use the tools to think like a CMO, not just a founder.
8. Simulate the market response
"Pitch this in a tweet. Now rewrite it as a skeptical CTO. Now as a testimonial from an enterprise sales lead."
Stress-test resonance before you ship.
9. Validate via search volume
"What keywords suggest intent for this enterprise AI automation tool? What’s the CPC?"
Build from demand, not just vibes.
10. Stack GTM prompts
"What’s the lowest-cost way to get 100 qualified users in health IT?"
Then: "Write a 5-email onboarding sequence."
11. Map the buyer journey
"What steps does a data science team take before buying a platform like this? What objections would procurement raise?"
Prime inputs for product, content, and sales. Whether you have survey results or access to Reddit or other relevant public forums, feeding data or pointing to online forums will always strengthen your analysis.
12. Ask for friction
"What makes this idea hard to build? Why might a CIO hesitate to buy it?"
Better to find the traps early, and having a MECE analysis can help you structure (and strengthen) your thinking.
13. Simulate a debate
"Argue for and against this startup’s success from the perspective of a founder, a healthcare buyer, and an enterprise investor."
Sharper than a SWOT analysis, and in my eyes, much more fun. Variation: pressure test your pitch with this.
14. Plan for your first 10 paying users
"Design a 10-day plan to get my first 10 enterprise customers for this workflow tool."
Momentum starts small — and fast. For this one, I’d encourage you to try asking for different approaches (SEO only, LinkedIn only, WOM only); or to list pros and cons of each approach; or both of these, and then (advanced mode) input your strengths as a team and ask which one it recommends.
15. Think two steps ahead
"If this AI-powered analytics layer succeeds, what’s true in 2 years that isn’t true today?"
This is a fun forcing function for pushing your thinking and long-term edge.
This isn’t just a productivity boost. It’s a shift in how strategy gets done.
Founders used to raise $2M to test an idea. Now? It might only take $20 and a prompt.
What once required a team, a roadmap, and a long runway now begins with conviction—and a chat window.
Prompt chains are emerging as a quiet edge. But most founders still aren’t using them. That’s the opportunity.
If you’re a solo founder, an operator exploring your next move, or a VC advising portfolio companies—this isn’t theory.
Before: Customer interviews, weeks of analysis, $5K in research.
After: One 15-minute chain with GPT-4, and a validated positioning map.
The infinite whiteboard is open. The feedback loop doesn’t sleep.
Got a prompt that helped you launch, test, or scale faster?
Drop it below. I’ll share the best ones in a future post.👇