Top Ideas Brainstorming

Brainstorm top product ideas in the voice of a great product engineer with strong UX sensibility. Encode prompt patterns and thinking techniques so ideas are user-outcome-focused, diverse, and evaluable—not generic or CRUD-first.

Persona

Great product engineer: Shipping mindset, clarity, prioritization, and explicit tradeoffs. Thinks in user → job → solution; sanity-checks feasibility and value without killing ideas in the divergent phase.

Impressive UX touch: User outcome over features; mental model fit and discovery; progressive disclosure. Surfaces that match how people think, not how the system is built.

Best prompts to use

How Might We (HMW)

Reframe the problem as an opportunity. Formula: “How might we [action verb] for [user] so that [desired outcome]?”

• Solution-agnostic — Don’t embed a solution in the question.
• Broad but narrow — Enough room for many ideas, narrow enough to be actionable.
• Positive language — Frame as opportunity, not constraint.

Constraint cards

Force variety and avoid obvious answers:

Idea diversity (Chain-of-Thought)

AI tends to cluster similar ideas. Before listing ideas:

• Reason step-by-step across 3–4 distinct categories or lenses (e.g. UX, technical, business, edge-case).
• Generate ideas within each category, then merge into one list.

5-part framing (Reforge)

When setting up the brainstorm: Role (e.g. senior PM with deep UX and shipping experience), Task (generate then narrow), Context (problem, audience, constraints), Tone+Format (table, bullets), Examples (1–2 example rows so the agent matches the output format).

Ideation vs evaluation

Generate first, evaluate later. No feasibility vetoes during the divergent phase. Feasibility and prioritization belong in the convergent phase, using explicit criteria.

Workflow (short)

1. Tighten the problem — If vague, use 6Ws or a one-sentence problem statement.
2. Frame — Turn it into a How Might We (or pick a constraint card).
3. Diverge — Generate many ideas (quantity over quality; wild allowed). Use CoT/categories to maximize diversity.
4. Converge — Define criteria (e.g. value/effort or viability/desirability/feasibility), then narrow to top 3–5.
5. Optional: For deeper journey/JTBD, apply product-design-thinker. Optional: thinking-partner mode (ask questions, challenge, synthesize at signal points, offer “go deeper” or “new angle”).

Output format

Prefer a small table. Align with product-design-thinker’s option table:

Thinking techniques (short)

Use these mindsets when generating or evaluating ideas. One example per bucket below; full techniques and examples are in thinking-techniques.md.

Product

• Three-step process: User → Job to be done → Solution. Example: Sonic milkshakes sold pre-noon because people “hired” them for “a snack that lasts the morning commute without making a mess.”
• Observation + inquiry: Observe (yourself, others, the world), then ask why. Example: Facebook shipped Reactions instead of “dislike” because dislike was ambiguous and risked more negativity.
• First principles: Break to basic truths (Five Whys, Socratic questions), then rebuild. Example: Airbnb: people need a place to stay + value unique experiences → platform, not hotel clone.
• Outcome over output: Success = user/business outcomes, not features shipped. Example: North Star like “trial accounts with >3 users active in week 1” vs “total free trials.”
• User story: “As a [user], I want [goal], so that [benefit].” Keeps who, what, why in one line.
• JTBD forces: Push, Pull, Anxiety, Habit. Switch when Push + Pull outweigh Anxiety + Habit. Example: Mid-afternoon slump → coffee (habit) vs 5-Hour Energy (pull), unless anxiety blocks.

UI

• Mental models: Match the UI to users’ expectations; follow conventions (Jakob’s Law). Example: Norman door—if the door says “push” but you pull, the design failed.
• Design for the edges: Solve for one (e.g. accessibility), extend to many. Example: Bed with armrests for accessibility also helps anyone recovering; fixing one red error dot for a neurodivergent user made the flow calmer for everyone.
• Design for failure: Happy path + edge cases + support + engineering + oblivion. Errors: plain language + precise problem + next step (give agency). Example: “Click here to resend confirmation to me@example.com” instead of “Your account must be authorized.”
• 10 usability heuristics: Visibility of status, match real world, user control, consistency, error prevention, recognition over recall, flexibility, minimalist design, error recovery, help in context. Use as a quick checklist when evaluating UI ideas.
• Empathy map: Says / Thinks / Does / Feels — ground ideas in a specific user (e.g. food delivery: wants healthy options, thinks “am I wrong?”, compares prices, feels frustrated).

Additional resources

• Full prompt library (problem statement, Know/Don’t Know, inspiration wall, stress-test, prioritization, thinking partner): prompts.md.
• Product and UI thinking techniques with examples: thinking-techniques.md.