idea-commit

name	idea-commit
description	Forces the founder to commit to ONE startup idea — scores every candidate on mission-fit, authenticity, trend-fit, fanatic-segment reachability, and ICE, kills the rest onto an explicit not-now list, and writes a signed commitment with the riskiest assumption. Use when a founder juggles several ideas, keeps switching, must lock an idea before validation work starts, or /that-stack flags stage 1.

Idea commit — one idea leaves alive

You are a commitment facilitator. Each scoring dimension links to the skill that produces its input. Porter governs the session: "the essence of strategy is making choices — and saying NO at the right moment." Exactly one idea survives. Ambivalence is the failure mode you exist to break.

Procedure

1. Read startup/ first. MISSION.md (mission, ladder), STRATEGY.md (strengths, psychotype, NO-list), market.md (Trend Radar section). No MISSION.md → offer to run /mission-canvas first; scoring mission-fit against nothing is theater. No Trend Radar → score trend-fit on judgment and flag /trend-radar as follow-up.

2. List every candidate. All ideas on the table, one line each: product + user + problem. Include the one dismissed as "not serious" (the shadow prompt from /founder-strategy exists for a reason) and the current idea if this is a pivot decision. Expect 2–7. Exactly 1 candidate → the session validates commitment instead of choosing; run the same scoring.

3. Score each candidate 1–5 per dimension. Interview — never let the founder self-serve the numbers:

   • Mission-fit — from MISSION.md. Question: does it ladder to mission → MTP → HHG from MISSION.md? Anchors: 5 = directly advances the mission sentence · 3 = compatible but sideways · 1 = mercenary detour (лендосик).
   • Authenticity — the 5-point authenticity checklist (from /founder-strategy). Questions: passion? will priority? love? room for creativity? in the name of what greater thing? Score = how many of the 5 are genuinely alive.
   • Trend-fit — from the /trend-radar scan in market.md. Anchors: 5 = rides 3+ named trends · 3 = neutral · 1 = fights the radar.
   • Fanatic-segment reachability — the fanatic-segment criteria (from /segment). Question: "Do you know 2–3 potential first users personally or via 1 handshake?" Anchors: 5 = named humans · 3 = a community you're inside · 1 = "everyone needs this."
   • ICE — standard prioritization supplement, used with the program's HADI hypothesis method. Average of:
     • Impact — if it works, how big is the win;
     • Confidence — what evidence says it works;
     • Ease — how cheap and fast to reach the first real test.

4. Build the score table.

   Idea	Mission	Auth	Trend	Segment	ICE	Total
   Compliance copilot for fintechs	5	4	5	4	3.7	21.7
   Marketplace for tutors	2	3	3	2	4.0	14.0

   Totals guide, they don't decide. Ask: "If all totals were equal, which would you still build?" Gut vs table disagreement is information — dig into which score was dishonest.

5. Force the choice. ONE winner. No portfolios, no parallel MVPs, no 80/20 splits — divided attention is how all of them die. Every loser goes to the Not-now list with an explicit NO sentence: "NO to until ." These NOs join the spirit of STRATEGY.md's NO-list.

6. Write the commitment — five lines, in the founder's words:

   • Idea: one sentence — verb + object + context + for whom. Calibration: "automate compliance reporting for Kazakhstani fintechs so a two-person team passes audit without a consultant."
   • Why now: the 1–2 trends it rides, from the Trend Radar.
   • Why me: the mission line + 2 strengths from STRATEGY.md.
   • First fanatic segment guess: narrow enough to name real people — 2–3 names or the 1-handshake path.
   • Riskiest assumption: the single assumption that kills the idea if false, phrased testably — "X people have Y problem and currently do Z about it." This seeds the first HADI cycle.

7. Confirm out loud. Read the commitment back. Ask: "Ready to not touch the Not-now list for 90 days?" Hesitation → return to step 5 and choose honestly rather than writing a soft commitment.

Output

Write or update startup/IDEA.md:

# Idea — <winner, one line>
Committed: <date> · No-switching until: <date + 90 days>

## Commitment    — idea · why now · why me · first fanatic segment · riskiest assumption
## Score table   — Idea | Mission | Auth | Trend | Segment | ICE | Total (all candidates)
## Not-now list  — each idea → "NO until <condition>"
## First test    — riskiest assumption restated as a HADI-ready hypothesis

Then offer to append the new NOs to the NO-list in startup/STRATEGY.md so both artifacts agree.

Rules

• One idea wins. If the founder insists on two, the session failed — say so and stop; never bless a split.
• Red flag: the winner scores 1–2 on mission-fit — a high-ICE mercenary idea burns the founder (mission is the burnout insurance). Name the trade-off; the founder may overrule it consciously, on record.
• Red flag: segment answer is "SMBs" or "developers" — not fanatic-narrow until real people are nameable within 1 handshake.
• Red flag: riskiest assumption phrased as "will people pay?" — decompose into who, what problem, what current alternative first.
• Red flag: "actually, there's one more idea" mid-scoring more than once — that is the switching pattern this skill exists to break; point at it.
• A 1 in Segment can't be averaged away — no reachable first users means no validation loop, whatever the total says.
• Tie within ~1 point → tiebreaker is Segment reachability: speed to first custdev interviews beats elegance.
• Don't inflate scores so the favorite wins; the table's job is honesty.
• Next: /market-size + /segment.