Startup Forge

• Treat a startup as a search for a business model: find the customers with urgent pain and a budget before building for them (Blank).
• Ask about past behavior, not hypotheticals, and tell evidence from compliments (Fitzpatrick, The Mom Test).
• Spend runway on experiments that test the riskiest assumption first (Ries; Bland & Osterwalder).
• Pivot when evidence demands it, and know the difference between a pivot and a panic.
• Read a term sheet: SAFE vs. priced round, the option pool, a 1x liquidation preference, and when to counter or walk away (Feld & Mendelson).
• Judge survival the way investors do: default alive or default dead (Graham).

1. 1
   Stage opens

   New stage = new deliverable to produce (problem brief, discovery synthesis, MVP plan, pivot decision, pitch deck), with a briefing that uses the team's real cash and month.
2. 2
   Interviews

   At Stage 2, teams interview AI customers, each from one of the industry's customer segments. Facts customers state count as evidence; warm answers to hypothetical questions are false positives.
3. 3
   Experiments

   At Stages 2-4, teams spend cash and a slot (two a stage, plus one per hire) on interviews, landing pages, pricing tests, concierge MVPs and paid pilots that return noisy readings of the real market.
4. 4
   Build the deliverable

   Team writes the actual artifact in-app, naming the segment it targets. The AI Critic scores it against a rubric, with the team's own evidence in view, up to three times per stage.
5. 5
   Stage 4 crisis

   Each team is dealt one industry crisis that changes the market. Persevere keeps burn flat, zoom-in cuts it, pivot costs a one-time bill; the decision pays off when the new target fits the team's evidence.
6. 6
   Stage close

   The class moves on: deliverables lock, months of burn come out of cash, and from Stage 4 each month earns the customers and revenue the team's target segment really gives it.
7. 7
   Demo Day (Stage 5)

   Three AI investors vote on a deck whose traction the sim verified. Two yes votes make the class's one fund an offer; after the partner meeting and the term-sheet decision, the fund backs only the best two or three teams.

Every team is scored on four pillars that compound across the sim. They aggregate into a composite the leaderboard ranks by, and the professor can see per-pillar deltas in the gradebook.

• A hidden market

  Each industry has three or four customer segments; only one has urgent pain and a budget, and one sounds excited but rarely buys. Customers, revenue and investor money all come from the sim, not from what a deck claims.

• AI customer interviews

  Customer personas who hold hidden truths and follow the Mom Test. The sim notes which facts a team uncovered and which answers were just compliments, and the professor can read the full log.

• Experiments that cost runway

  Interviews, landing pages, pricing tests, concierge MVPs and paid pilots, each with a cost and noisy results. Hires add capacity. Testing the riskiest assumption first wins.

• Demo Day with one fund

  Three AI investors vote on a deck with sim-verified traction, the partner meeting follows, and a class fund backs only the best two or three teams, at terms set by evidence rather than typed valuations.

• AI Critic

  Scores every deliverable against a stage rubric, with the team's earlier stages, its industry's norms, its real numbers and its own evidence in view. Gets stricter in Death March.

• 5 business types

  B2B SaaS, DTC Products, Consumer Mobile, Marketplace, AI Product: same engine, different customer segments, crises and investor norms.

• Light founder roles

  CEO, Customer Lead, Product Lead and CFO, each owning part of the recorded work (deliverables, interviews, experiments, the term-sheet answer), with a short reflection from each student after every stage that the professor reads on one page.