Agents in Stateful

Stateful can run agents for you. Switch to the Agents view (the Context / Agents toggle at the top of the app), describe what you want in a sentence, and Stateful turns it into an agent you can review and activate.

Agents are in beta, like the rest of Stateful.

What an agent is

An agent is a small, recurring job with memory. It:

1. Observes your data — email, calendar, todos, contacts, calls, chats, files, your wiki, and the web.
2. Thinks using your context, the same way the assistant does.
3. Reports back through a channel you choose.

Agents are read-only over your data by design. They watch and summarize — they don't send email as you or change your files. Think "morning briefing," not "autopilot."

Building one

1. Open the Agents view and describe the agent in plain language — e.g. "Each morning, summarize anything urgent in my inbox and flag meetings I'm unprepared for."
2. Stateful compiles that into a draft — a name, a prompt, and the tools it's allowed to use.
3. Review the draft, adjust anything, and activate it.

Templates are tailored to your workspace (solo, household, or team), so you can start from a relevant example instead of a blank page.

When it runs

Choose a trigger when you activate the agent:

• On a schedule — every few minutes, daily at a set time, or a custom cadence.
• Before calendar events — so a prep brief lands ahead of a meeting.
• On Claude Code events — so an agent can react while you're coding.

How it reports

An agent can deliver through in-app chat, email, SMS, push notification, or a webhook. Every run is logged, so you can open the agent and read its history.

Sharing

You can share an agent — as a recipe (the setup, for someone to run on their own data) or in full — and clone agents shared with you.