Each prompt below is designed to feed into Claude or ChatGPT. Run them in order — outputs of earlier prompts feed into later ones. Prompts 1-3 are the scope cut — the most important and the most skipped. Don't move to prompt 4 until you've genuinely cut something that hurt to remove.

1. Feature autopsy. "Here's the model I priced in M3 and the frame from M4: [paste]. Here is everything I currently think the MVP needs to do, in one long list — features, pages, integrations, content, polish: [paste]. For each item, classify as: (a) the customer pays specifically for this, (b) the customer expects this as table stakes, (c) the customer would be fine without it for v1, (d) I want this because it makes me feel like a real founder. Be brutal in category (d). Output the four lists separately."
2. The 20% that earns 80%. "From the lists above, identify the smallest possible bundle that satisfies category (a) and the minimum viable level of category (b). Strip every (c) and every (d). State the bundle in one paragraph — what the customer actually gets, and what they explicitly don't get. If the bundle takes more than 30 days to ship with AI tooling, cut again. Tell me what I'm cutting and what it costs me to cut it."
3. The 'no' list. "Write the explicit not-in-v1 list — every feature, page, integration and polish item I'm deferring. For each, write the trigger that would put it back on the list (e.g. '10 customers ask for it', '3 customers refuse to pay without it', 'revenue passes $X'). This is the list I look at when I'm tempted to add something back in week 2."
4. Build stack — what AI does, what I do. "My MVP scope is [paste prompt 2 bundle]. My skill base is [paste — what I can / can't build]. Map the build into stack layers: (a) content / copy / brand voice, (b) interface / front-end, (c) logic / back-end / automation, (d) payments + delivery, (e) ops / fulfilment. For each layer, recommend the AI-native tool I should use (Claude / GPT / Lovable / Cursor / v0 / Webflow / Shopify / Notion / Zapier / Make / etc.), the realistic time to build with that tool, and the one human decision in that layer that AI shouldn't make for me. Bias toward fewer tools — every integration is a tax."
5. 30-day ship plan, working backwards. "Today's date is [paste]. Ship date is 30 days from today. Work backwards: by day 30 the MVP is live and 3-5 customers have paid. By day 25 the first buyers are in. By day 20 the build is functionally complete and being tested. By day 15 the spine is up. Give me a week-by-week plan: week 1 outcomes, week 2 outcomes, week 3 outcomes, week 4 outcomes. Each outcome must be a demonstrable artefact, not 'work on X'. Flag the single hardest week."
6. First-5-buyers list. "From M2 I interviewed 15 people in my segment. From that pool, plus any warm contacts inside my network, list 12-15 named humans who are the most likely first buyers — not 'people like them', actual names. Rank by (a) intent signal from the M2 interview, (b) ease of reaching, (c) likelihood of giving honest feedback after they buy. Top 5 are who I message in week 1. Bottom of the list are who I burn last. If I don't have 12-15 names, the gap is the work — go back and find them."
7. The first message. "Write the first message I send to each of the top 5, in my voice. Not a launch announcement. A personal note that names what I built, why I built it for them specifically (reference their M2 interview), the price, the exact thing they get, and what I'm asking them to do in the next 5 minutes. Two versions: warm (we've spoken) and lukewarm (we've spoken once, briefly). No 'launching today!' energy. No emojis unless I always use them. They are buying a product, not joining a movement."
8. The feedback loop — what to ask, what to ignore. "For each of the first 3-5 buyers, design a 15-minute post-purchase conversation I run within 7 days of their purchase. Five questions, exact wording. Cover: (a) what specifically made them say yes, (b) what almost made them say no, (c) what they expected that didn't show up, (d) what showed up that they didn't expect, (e) what they would tell a peer about this if asked tomorrow. Then tell me which categories of feedback I act on immediately, which I log for v2, and which I ignore on purpose at this stage. The hardest skill is ignoring the wrong feedback."
9. The scope-creep firewall. "Based on my model [paste M3] and frame [paste M4], list the 5 most likely scope-creep traps in the next 30 days — the features, pages or 'just one more thing' moments that have killed similar MVPs. For each: the trigger (when it tends to surface), the cost (days lost to ship date), and the exact answer I give myself or my team when it surfaces. The goal is to have the 'no' rehearsed before the temptation lands."
10. The one-page MVP brief. "Write my one-page MVP brief, suitable to print and pin next to the founder frame from M4. Sections: (1) what the MVP is, in one sentence a buyer would understand, (2) what's in v1 and what's explicitly not, (3) the build stack — tool by layer, (4) the 30-day ship plan with the four week-by-week outcomes, (5) the named first 5 buyers and the message I'm sending each, (6) the feedback questions and the ship-day-plus-30 review date. Plain English. No jargon. Fits on one A4 page. The page I look at the day I want to add something."