Persist Now, Model Later
Capture agent context today while your target schema evolves.
SELECT record_event('agent:summarized', ...); Open Source / PostgreSQL
Your CRM, your tasks, your agents — one graph you actually own.
For teams who want operational data and agent memory in their own Postgres, not scattered across SaaS platforms.
Rye is a flexible graph schema that handles contacts, tasks, events, and agent context in one place — your PostgreSQL database. No platform migration. No vendor lock-in. Start capturing context now and evolve your data model as you learn.
Works With Existing Data
Overlay on current systems of record. No migration required.
SELECT track_table('public','contacts'); Your Database, Your Rules
Data never leaves your Postgres. Row-level security enforces access on every query.
SELECT agent_node_summary(node_id); How It Works
Four steps from install to production agent queries.
- Gain visibility — install Rye and inspect your scope.
SELECT rye_catalog(); - Connect your data — link domain rows and enable change capture.
link_record(...)+track_table(...) - Preserve history — record activity and evolve assertions over time.
record_event(...)+supersede_assertion(...) - Query from agents — compact context and current state, ready for LLM consumption.
agent_node_summary(...)+current_assertions
What You Get
- No migration of source systems.
- Temporal history with auditable, supersedable assertions.
- Agent-compatible read and write patterns with PostgreSQL controls.
- Row-level security enforced on every table, every query.
- Content-addressed artifact storage with deduplication.
Start Here
- Core Contract Guarantees Rye makes and expects
- Data Dictionary Every table, column, and trigger explained
- Security Model RLS policies, redaction, and access control
- Quickstart Working system in under 10 minutes
Ready to start?
Run the quickstart to have a working Rye instance in minutes, or read the overview to understand the architecture first.