Stateful docs
Building agents

How agents use Stateful

The two ways agents and Stateful work together.

"Agent" can mean two things with Stateful. Both rest on the same idea — Stateful is the memory layer — but you set them up differently.

Agents you build in Stateful

Stateful has its own Agents surface: you describe what you want in a sentence, Stateful turns it into an agent, and it runs on a schedule or trigger — observing your data and reporting back. These agents are read-only over your data: they watch and summarize, they don't act as you. See Agents in Stateful.

External agents that use Stateful

The AI tools you already use — Claude.ai, Claude Code, Cursor, or an agent you build yourself — can read and write your Stateful context over MCP. Stateful becomes their long-term memory, so they start every run grounded in your real state. See Bring your own agent.

Whichever path you take, the memory lives in Stateful — not in the tool — so it persists across sessions and across tools.

On this page

How agents use Stateful