Three critics for every flow. Run accessibility, edge cases, and business alignment in one pass before you ship.

When I'm about to sign off on a flow, I don't ask AI for generic feedback. I ask it to review from three fixed perspectives.

Same flow. Three critics. One synthesized output.

Review this flow from three perspectives:

Critic 1: Accessibility
Critic 2: Edge cases
Critic 3: Business alignment

Present all findings together with recommendations.

Each critic catches different failure modes. Accessibility surfaces keyboard traps and unclear labels. Edge cases surfaces empty states, errors, and weird inputs. Business alignment surfaces whether the flow actually moves the metric or policy it claims to support.

Paste the flow (screens, prototype link, or spec). Run the prompt. Fix what's real before polish.