Most founders use AI to write emails faster. A small number use it to stress-test their assumptions before a $500K decision. The gap between those two use cases is mostly just prompt quality.
The Problem with How Founders Use AI
The default founder workflow looks like this: open ChatGPT, type "write me a cold email to VCs," get something generic, ship it. The output is fine. The thinking is untouched.
That's the issue. AI's highest-leverage function for a founder isn't content generation — it's structured adversarial thinking. You need a sparring partner who knows your market, never gets tired, and has no social incentive to tell you what you want to hear. That's what these prompts are trying to approximate.
The prompts below aren't magic incantations. They're frameworks — questions designed to surface the information you're actively avoiding. Each one offloads a specific type of cognitive work: assumption auditing, competitive pressure-testing, narrative stress-testing, decision structuring. Use them as starting points and iterate toward your context.
One caveat before we start: AI is only as useful as the context you give it. Paste in your deck, your docs, your financials where relevant. Prompts without context are just therapy with extra steps.
The 10 Prompts
1. The Pre-Mortem
The prompt:
I'm about to [launch a product / make a hire / close a deal / raise a round].
Assume it's 18 months from now and this decision turned out to be a serious mistake.
Walk me through the 5 most likely reasons it failed.
Be specific — generic failure modes don't count.
Why it works: The pre-mortem is a well-documented technique from Gary Klein's research on decision-making. By assuming failure in advance, you bypass the optimism bias that kills most startup planning. The key constraint — "be specific" — forces the model past "bad product-market fit" into something actionable like "you launched in Q4 when your target buyer had frozen procurement budgets."
2. The Devil's Advocate Investor
The prompt:
Here is my pitch deck / one-pager / product thesis: [paste content].
You are a skeptical Series A partner at a top-tier VC. You've seen this exact category before and it underperformed. Walk me through your 7 hardest questions — the ones you'd ask to disprove the thesis, not validate it. Then tell me which answer would actually change your mind.
Why it works: Most pitch prep is defensive: you anticipate easy questions and prepare polished answers. This flips it. The "what would change your mind" clause forces the model to identify the single load-bearing assumption in your story — which is almost always the thing you haven't figured out yet.
3. The Competitive Landscape Autopsy
The prompt:
Here are 3-5 companies in my space that failed in the last 5 years: [list them].
For each one, give me your best analysis of the actual reason they failed — not the press release version.
Then identify the pattern that connects them.
Finally, tell me which of those failure modes applies to my company: [describe your company].
Why it works: Startup failure analysis is usually self-serving. Founders study competitors' deaths to feel better about their own survival odds. This prompt inverts that. You're looking for the recurring structural problem in your category — and then applying it inward.
4. The Persona Interrogation
The prompt:
You are [specific job title] at a [specific company type/size] dealing with [specific problem].
I'm going to pitch you my product. Your job is to push back.
You're not hostile, but you have three existing solutions, a tight budget, and a boss who's skeptical of new tools.
Ask me the questions a real buyer would ask — including the ones that would kill the deal.
Why it works: Customer discovery is chronically soft because real customers are polite. This gives you a consistent, repeatable buyer persona that will actually challenge your assumptions. The more specific you make the persona, the harder the questions get. "VP of Engineering at a 200-person SaaS company whose team just botched a migration" will give you better pushback than "a software leader."
5. The Minimum Viable Narrative
The prompt:
Here is everything I know about my company: [paste your deck, docs, or a brain dump].
Distill this into the single most compelling 3-sentence pitch — no jargon, no product features, just the problem, the insight, and why we win.
Then give me 3 alternative versions of that pitch targeting different audiences:
(a) a technical co-founder candidate, (b) a non-technical enterprise buyer, (c) a generalist angel investor.
Why it works: If you can't produce a clean 3-sentence version, your thesis isn't clear — not to audiences, and probably not to yourself. The multi-audience exercise forces you to separate the core insight from its packaging, which is a useful muscle for anyone who's been heads-down building for six months.
6. The Pricing Stress Test
The prompt:
My product currently costs [X] per [month/seat/usage unit].
My target customer is [describe them].
Give me the strongest argument that I'm underpriced by 3x.
Then give me the strongest argument that I'm overpriced and will lose deals at renewal.
Finally, recommend what data I'd need to collect in the next 90 days to know which argument is right.
Why it works: Pricing decisions at early-stage companies are almost always made with insufficient data and too much founder intuition. This prompt doesn't tell you what to price — it identifies the crux of the uncertainty and gives you a research agenda.
7. The Hiring Spec Auditor
The prompt:
Here is a job description I'm about to post: [paste JD].
Tell me:
(a) What type of person this will actually attract vs. who I probably want
(b) Which requirements are proxies for things I actually care about — and what those things are
(c) What a strong candidate would find off-putting about this posting
(d) One sentence that would make a top-tier candidate stop scrolling
Why it works: Most job descriptions are written by founders who copy a template and add their own buzzwords. The result attracts the median candidate. This prompt surfaces the gap between what you wrote and what you meant — which is usually significant.
8. The Strategic Option Tree
The prompt:
I'm facing this decision: [describe the decision and its context].
Map out the 4-5 most meaningfully different paths I could take — not variations on the same path, but genuinely different strategic bets.
For each one, tell me: what it's optimizing for, what it's sacrificing, and what early signal in the next 60 days would tell me it's working or not.
Why it works: Under pressure, founders collapse decision spaces. "Should we pivot?" feels binary. This prompt forces the option space open before you close it, and the 60-day signal question anchors each option to something falsifiable.
9. The Investor Update Critic
The prompt:
Here is my investor update draft: [paste update].
Read it as a board member who is quietly worried about the company and looking for signals that the founder either has clear-eyed awareness of the problems or is papering over them.
Tell me:
(a) What I buried that I should have led with
(b) What the update signals about my mental model of the business
(c) One thing I should have included that would have increased confidence
Why it works: Investor updates are rarely written to communicate — they're written to manage perception. The board member framing gets you honest feedback on the subtext, not just the content.
10. The Second-Order Consequences Audit
The prompt:
I'm planning to make this decision: [describe it — a product change, a partnership, a pricing shift, a team restructure].
Walk me through the second and third-order consequences — the effects that don't show up immediately, that will surprise me in 6-12 months if I haven't thought about them now.
Focus especially on: team dynamics, customer behavior, competitive response, and technical debt.
Why it works: First-order thinking is easy. You add a freemium tier → you get more users. Second-order: your support burden triples, your paid conversion rate drops as customers wait for free features, and your best enterprise prospects now associate your brand with "free." This prompt is a cheap way to simulate the consequences you'll pay for later.
Why These Work (and When They Don't)
The pattern across all 10 prompts is the same: they force the model into a role with a specific adversarial or analytical posture, constrain the output to be specific rather than generic, and attach the output to a concrete decision or artifact.
Generic prompts get generic outputs. "What are the challenges of building a B2B SaaS startup?" will give you a business school lecture. "You're a skeptical Series A investor and here's my deck — what kills the deal?" gives you something you can actually use.
The failure mode is confirmation-seeking. If you prompt for validation ("what's good about my pitch?"), you'll get a cheerleader. The prompts above are designed to fight that instinct. Resist the urge to soften them.
One structural limitation worth naming: AI doesn't know your specific market the way a domain expert does. These prompts surface frameworks and pressure-test logic — they don't replace a 30-minute call with someone who's operated in your exact category. Use them to prepare for that call, not to skip it.
The Takeaway
- Specificity multiplies value: The more context and constraints you add, the more useful the output. "Be specific — generic failure modes don't count" is doing real work in these prompts.
- Adversarial framing beats neutral framing: Assign the model a role with a reason to push back. Skeptical investor, worried board member, disengaged buyer — these personas generate better questions than open-ended ones.
- Attach outputs to decisions: Prompts that end with "what data would I need" or "what signal in 60 days" convert analysis into action. Prompts that don't are just interesting.
- Use these iteratively: The first output is a draft, not a verdict. Push back on the model's reasoning. Ask it to steelman the counterargument. The second and third turns are usually more valuable than the first.
- Context is load-bearing: Paste in your actual documents. A prompt about your pitch deck without the deck is a prompt about nothing.
Tags: prompting, startups, AI, productivity, founder-tools