# LOG PROJECT INSIGHTS PROCESS (Data-Centric, Structured Interaction, AI Assists) ## Process Steps: 1. AI: Verify Schema access. Access Project State. 2. AI: Initialize/Access Insights Log in `project_state.logs.insights`. 3. AI: Receive user input (insight details) or identify insight internally. 4. AI: Parse input/context and propose a structured `insight_object` (draft). 5. AI: Formulate yes/no questions to confirm each field of the proposed insight object. 6. AI: Present proposed insight and questions to user. 7. User: Provides yes/no answers. If "No"/freeform correction, AI refines that part of the proposal (back to step 4) and re-confirms. 8. AI: Update draft `insight_object` based on "Yes" responses. Set status to 'Logged'. 9. AI: Add finalized `insight_object` to `project_state.logs.insights`. Update metadata timestamp. 10. AI: Ask "Insight logged. Log another insight? (Yes/No)". 11. AI: If "Yes", return to step 3. If "No", signal completion. ## Example Structured Questions (AI to formulate): * "I understand the insight to be: '[Proposed description]'. Is this correct? (Yes/No)" * "Does this insight stem from [Proposed source, e.g., 'Task X']? (Yes/No - If No, please specify or state 'General')" * **[IF a proposal was rejected]:** "Understood. Please clarify the [Field Name, e.g., description of the insight]." (Freeform input expected) * **[IF insight is logged]:** "Insight logged. Log another insight? (Yes/No)" ---