Thoughts

5 thoughts about "Open Brain" in the last 30 days
4/1/2026

[Session Log] Notion → Open Brain bridge: Built and tested the Notion sync pipeline. Queried Notion Projects database (ID: 2ced1125-90e3-8162-ae59-d5720c2b86ff) and successfully pushed 2 test projects into Open Brain. Set up a local Cowork scheduled task (daily 4am) and saved cloud task instructions at reference/instructions/notion-sync-cloud-task.md for future migration to Claude Code cloud scheduled tasks. Key finding: TATTOO HQ page uses linked database views — must query source databases directly (Projects DB: 2ced1125-90e3-8162-ae59-d5720c2b86ff, Intake Submissions DB: 2cfd1125-90e3-8097-8aae-db0565f33a77). This completes capture point 8 of 9 for Open Brain. Remaining: Instagram signals.

Google Apps Script triggers don't require "Deploy." Save the script + set a time-based trigger is sufficient. Deploy is only needed to expose the script as a web app or API endpoint. This confused Dave during the Open Brain calendar and Gmail bridge setup.

People: Dave

Open Brain passively listens to Dave's existing Second Brain Slack channel, silently embedding every message into the vector store in the background without interfering with the existing Notion pipeline. This means everything Dave sends to Second Brain also gets vector-indexed in Open Brain for semantic search, while Notion remains the system of record for structured data.

People: Dave

The relationship between Dave's Second Brain and Open Brain: Second Brain is the structured project manager (Slack to Notion, handles tasks/projects/clients/payments). Open Brain is the semantic memory layer underneath it (Supabase pgvector, fuzzy long-term memory). Dave built Open Brain because he was self-censoring his Slack captures, not wanting to clutter Notion with half-baked ideas and fleeting observations. Open Brain catches everything Second Brain doesn't, and any AI agent can search and write to it across platforms and sessions.

People: Dave

Phase 3 of Open Brain is being deployed. Added HMAC verification, two-channel routing (capture channel + silent Second Brain listener), and query detection via search:/ask: prefixes.