Can I use Myco and Letta together?
Yes. Many teams can use Letta for agent runtime and Myco as the persistent memory backend.
Comparison
Myco vs Letta: Agent Runtime Memory vs Organization Memory Layer
Letta is excellent for agent runtime orchestration. Myco is purpose-built as the persistent memory substrate teams share across agents.
| Feature | Myco | Letta |
|---|---|---|
| Primary scope | Knowledge substrate | Agent runtime |
| Deterministic write path | Yes | Not a stated focus |
| Workspace governance | Native | Not a stated focus |
| Cross-agent shared memory | Native Postgres source of truth | Custom |
| Source-linked provenance | First-class via brain_why | Partial |
| Import ChatGPT/Claude history | mycobrain-ingest --from chatgpt/claude-export | — |
| Read-only REST for non-MCP consumers | mycobrain-rest (search + why, loopback) | — |
| Operational mode | Self-hosted (Apache-2.0); managed cloud waitlist | Open runtime |
Bottom line
Use Letta when you need runtime agent loops. Use Myco when you need durable shared memory, lineage, and governance across your stack.
FAQ
Does Myco support workspace isolation?
Yes, per workspace. Every agent authenticates with a workspace- and agent-scoped API key, and isolation is enforced by Postgres RLS under the least-privilege role the agency kit ships; the default local quickstart runs as a superuser and bypasses RLS.