capx-ai/casa/casa-ideate

casa-ideate

The option-generating craft. Quality comes from explicit rejection, not optimistic ranking. Generate widely, kill most, explain the few that survive.

Steps

1. If company-brain/profile.json does not exist, tell the founder to run /casa-start

first and stop.

1. Ground the question. Read company-brain/STRATEGY.md, profile.json,

build-map.json, the latest decisions/, and learnings.jsonl. Frame the specific decision the founder faces and the constraint it lives under (budget, stage, channel, and any token or on-chain layer if that is in scope).

1. Generate many, deliberately varied. Produce eight to twelve options spanning real

different angles (cheap and fast, high-ceiling and risky, contrarian, adjacent to what is working). Each grounded in the actual company state, not generic startup advice. No founder-bro framing.

1. Critique all of them. For every option, state the strongest reason it fails: weak

evidence, wrong stage, no distribution, poor unit economics, off-strategy, already tried (check learnings.jsonl). Rejection with a reason is the mechanism.

1. Surface the survivors. Keep the two or three that withstand their own strongest

critique. For each, give the why-it-survives, the cheapest test, and the metric it moves. List the rejected ones in one line each so the founder sees the field.

1. Route the winner. On the founder's pick, hand to a brief or directly to

casa-build for the playbook that executes it. Append a note to decisions/ so a rejected idea is not re-litigated later.

Rules

• Explicit rejection is required. An option with no stated failure mode was not

critiqued.

• Ground every option in this company's real state and past learnings, not generic

advice.

• The founder chooses. You generate, critique, and recommend; you do not commit.
• No em-dashes, no emojis. Institutional tone.