Relmem

Relmem is an AI relationship memory layer for founder-led teams. It ingests Gmail, Slack, LinkedIn, Attio, Granola, WhatsApp, and Google Calendar touchpoints, stores workspace-scoped contact memory in PostgreSQL and Qdrant, detects stale relationships, drafts follow-ups with Claude, and uses Slack plus a small ops dashboard as the UI.

Run locally

1. Copy .env.example to .env and fill in real API keys.
2. Start services:

docker compose up --build

3. API health check:

curl http://localhost:8000/health

MVP endpoints

• POST /auth/gmail/callback
• POST /auth/slack/callback
• POST /webhooks/granola
• POST /webhooks/whatsapp
• GET /webhooks/whatsapp/status
• GET /contacts
• GET /contacts/{id}/history
• POST /contacts/{id}/approve
• POST /ingestion/sync
• POST /billing/checkout
• POST /billing/webhook
• POST /slack/events

Constraints implemented

• Gmail, Attio, WhatsApp, and Google Calendar have concrete ingestion paths; Slack, LinkedIn, and Granola hooks are scaffolded.
• Qdrant retrieval requires workspace_id and contact_id and uses payload filters.
• Claude prompt context includes platform source, timestamp, and content for every touchpoint.
• Attio writeback only runs from approval endpoints or Slack approval actions.
• Touchpoint chunking uses a 512 token max with 50 token overlap.