Replacing Heuristic Checklists with Adaptive AI: A UX Case Study at Forma

AI · 6 min read

Forma, a mid-stage design platform, found teams spent excessive time iterating on the same heuristic checklist items during reviews. Manual checklists were rigid and produced fatigue, while designers reported checklist-itis where boxes were ticked without deep consideration. The product team proposed an AI assistant that would suggest heuristics and prioritize them based on project context and historical fixes.

The AI model was trained on anonymized review notes, ticket outcomes, and final release changes to predict which heuristic violations correlated with production issues. UX decisions focused on transparency: suggestions included confidence scores, example screenshots, and a clear explanation of why a suggestion mattered. Designers could accept, dismiss, or annotate suggestions, creating a feedback loop to retrain the model.

In a controlled pilot across 10 teams, review time per feature dropped by 22% and the percentage of accepted suggestions stabilized at 48%. Qualitative feedback showed designers appreciated contextual recommendations but demanded better control, so the team added preference controls and an opt-out for sensitive projects. The case highlights that AI is most effective when it augments decision-making and preserves human judgment rather than replacing it.