The Founder's Claude Desk
"Claude prompts every founder should try" — two dozen of them, in a grid. The prompts are genuinely good; the format wastes them. Twenty-four separate one-liners are a menu you scroll past, not a system you use. We grouped them into six jobs a founder actually returns to, made each tool-neutral, and attached the one thing a prompt grid never includes: how to check the answer before you bet on it.
01 — The reframeSix jobs, not twenty-four prompts
The prompts cluster cleanly. Strategic advisor, decision partner and investor- perspective are one job: pressure-test my thinking. Persona builder, competitor analysis and market strategy are one job: understand the market. The founder's week has maybe six recurring questions; here they are, each as a standing tool rather than a prompt to re-find.
1 · Pressure-test the decision
Strategic advisor, devil's-advocate, investor-perspective, decision partner. Feed the decision, the context, and what you're afraid of; ask for the case against, the blind spots, and a recommendation with its reasoning.
VerifyA model will argue any side. Ask it to argue the opposite next, and trust the decision only if it survives both.2 · Understand the market
Customer personas, competitor analysis, go-to-market, market trends. Give it your real customer notes and competitor URLs, not hypotheticals.
VerifyModels infer market facts they can't know. Every competitor claim gets checked against the competitor's actual site before it enters a plan.3 · Sharpen the message
Website-copy audit, sales-email sequence, pitch-deck improvement, content strategy. Best used with your current copy pasted in, plus who it's for.
VerifyRead it aloud in your voice. Generated marketing copy has a detectable default register (see I-10) — cut anything that sounds like everyone.4 · Run the operation
SOP creator, onboarding plan, hiring scorecard, meeting-summary-to- actions, workflow optimization. This is where the durable weekly value lives.
VerifyAn SOP is a hypothesis until someone follows it once. Pilot it with one person before it becomes policy.5 · Plan and prioritize
90-day growth plan, product roadmap, growth-experiment generator, pricing-strategy review. Anchor to your real constraints — runway, headcount, one goal.
Verify"25 growth experiments" is easy; ranking by expected impact against your actual capacity is the work. Cut the list to the three you'll truly run.6 · Report and reflect
CEO dashboard, investor update, customer-feedback analysis, AI-automation opportunities. Same template every period so trends surface.
VerifyAn investor update built from generated optimism erodes trust fast. Every number traces to a source; every "win" is one you'd defend live.02 — Why these prompts surviveThe genre is durable
Unlike product walkthroughs, role-and-task prompts like these don't decay with the next release — they're about framing your own thinking, which is why the pattern holds across model generations (the durable class in I-10). Made permanent, this desk is a Claude Project with your business context loaded once; each family becomes a saved prompt or a Skill — exactly the build order in C-10. The prompts are the seed; the system is the plant.
Every prompt here produces confident output. Confidence is not correctness. The verify line under each family is not optional garnish — it is the difference between a founder who uses AI to think faster and one who uses it to be wrong faster.