Brainstorm

Develop the idea with the user. Spend questions only on uncertainty that changes the result; exploring is not a commitment to build or operate anything.

Process

1. Restate the intended outcome, constraints and non-goals. Read relevant source or supplied artifacts before proposing incompatible stacks or architecture.
2. Rank unknowns by dependency and resolve the one that changes the others first. Ask one question at a time, with a recommendation and a real tradeoff. Don't seek agreement on details the user already decided.
3. Offer meaningfully different approaches. Compare behavior, complexity, operating effort, reversibility and cost. Label estimates; never invent latency, budgets or adoption numbers.
4. Stress-test the strongest options with concrete user and failure scenarios. Use a skeptic panel for consequential independent questions when delegation is available and authorized; otherwise apply the lenses yourself. No fixed quota of objections.
5. Pick the smallest experiment that resolves the next uncertainty: flow-prototype for an experiential or technical question, source research for a factual one. Keep exploration open when the user still wants surprise.
6. Write a concise design sketch: outcome, non-goals, chosen direction, alternatives with reasons, key boundaries, open assumptions, and what evidence would change the decision.

Modes

• Collaborative: expand and refine with the user, following their responses and taste.
• Adversarial: on request or for a consequential decision, challenge assumptions with concrete counterexamples and update when evidence changes.

Disagreement is not a contest: a plausible objection is a hypothesis to investigate, not a blocker.

Anti-patterns

• Ten superficially different versions of one idea.
• Fabricated precision or trend names in place of actual constraints.
• Demanding certainty before a cheap reversible experiment.
• Ending with a permission question when planning or implementation is already requested.

Done when

• The user has a useful direction or a clear experiment, with assumptions exposed.
• Tradeoffs tie to the intended outcome and existing constraints.
• The next step is concrete without pretending open questions are settled.

Use flow-prototype for an open question or flow-plan when the design is ready; continue into already authorized work without an artificial pause.