Brainstorming
Generating options before committing to an approach
BMAD Brainstorming is a structured diverge-then-converge session where multiple personas generate solution options before the team commits to any one approach. It prevents premature lock-in and surfaces alternatives the original author did not consider.
The most common failure mode in AI-driven development is committing to the first plausible solution. The AI produces a coherent approach, it looks reasonable, and implementation begins — but no one checked whether a simpler option existed, or whether a known constraint would have ruled out the chosen approach. BMAD Brainstorming inserts a structured options-generation step before any approach is locked in.
The session runs in two phases. In the diverge phase, each relevant persona generates one or more alternative approaches independently — without critiquing the others. The Architect generates technical options; the PM generates story-level framings; the Developer generates implementation paths. The constraint is: during diverge, all options are valid, no option is rejected, and the goal is quantity over quality.
In the converge phase, the team evaluates the generated options against the known constraints: technical feasibility, scope, dependencies, and time. The output is not necessarily the best option — it is the chosen option with a documented rationale for why the alternatives were rejected. That rationale is what makes the decision reviewable and reversible if constraints change.
Key Concepts
- Diverge before converge — no critique during options generation
- Each persona contributes options from their domain perspective
- Quantity over quality in the diverge phase — premature filtering kills good ideas
- Converge produces a chosen approach AND documented rationale for rejections
- Rejection rationale is required — it makes decisions reversible when constraints change