AI Team & Loops

Once an app is built, put AI to work operating it — not just building it once and walking away.

Staffing an AI Team

Every app can be staffed with a recommended team of agents that operate it day to day:

  • Handle the inbox — triage messages, orders, or support requests that come into the app
  • Propose actions — draft replies, flag issues, suggest next steps
  • Approval-gated — agents propose; you approve before anything user-facing happens (or auto-approve low-risk actions, depending on how you configure it)
1

List Your Apps

2

See every app you've built, with how many agents are staffed and how many items need your approval.

3

Hire a Starter Team

4

Pick a recommended team template for the app's type — an e-commerce app and a booking app get different starters.

5

Review What They Propose

6

New proposals show up for approval. Approve, reject, or let low-risk ones auto-resolve.

Autonomous Loops

A loop is a goal you hand to an app's hired agent — it advances the goal one step at a time, on a cadence you set, until the goal is met or it hits a step limit:

"Get my first 10 signups" "Grow weekly active users by 20%"

Each step is approval-gated the same way as regular proposals — a loop doesn't take irreversible action without your sign-off.

Custom Agents

Beyond app-staffing, you can also build a standalone custom AI agent — its own persona, system prompt, and model — for use cases that aren't tied to a specific app (a support chatbot, an internal assistant). Custom agents support persistent memory, file-based knowledge (RAG), scheduled tasks, and can be embedded on any website.

Using It

All of this is available from the app dashboard, the mobile app's Team/Inbox tabs, and as tools in the MCP integration (list_apps, staff_app, start_loop, list_loops, create_agent, chat_with_agent) — so ChatGPT, Claude, or Gemini can staff and run your apps directly.