Autonomous agents that manage your screens(and more)

A Screen Agent is a user-created worker you install on office screens. You give it sources (read access to tools like Slack or Notion), curation rules (standing intents for what to show), and optional skills. It keeps an AI-curated playlist fresh and fires triggers when something matters now.

Playlists play what a person scheduled.

Triggers switch content when a single defined event fires.

A Screen Agent sits above both: it holds ongoing curation rules, reads live sources on a refresh cadence (a “tick”), updates its own playlist for the loop, and chooses when to fire a trigger for moments that should interrupt.

Saved intents belonging to an agent: the prompt, plus the sources it may read. They say what to show, not how. The agent decides whether an item stays in the playlist loop or interrupts as a trigger.

Example: “When a Linear issue is blocked or ships, show it. Pull decisions from #product. Keep today’s meetings from Google Calendar on the board, and update the screen when the Notion launch page changes.”

Yes. By default an agent can use what is in the account: media you uploaded, dashboards you built, and Design Studio work. Scope points it at a folder, or a set of that content, so you emphasise what it should draw from.

It still curates, and it still reads the connected tools. Scope is how you keep control without turning the agent into a playlist you have to rebuild by hand.

Sources are the tools the agent reads, usually through MCP. Slack, Notion, and Google Calendar are the ones this page walks through, with Linear, GitHub, and HubSpot in flight. The ring of logos also includes tools we expect to add.

If one you rely on is missing, tell us. We add sources based on what customers actually use, not a fixed catalogue. Every source is read-only. Screens are shared spaces.

You set the schedule. A tick is one pass: the agent re-reads its sources and updates its one AI-curated playlist only when the composed content changed. It does not publish a new playlist every time it runs.

Yes. Some moments should not wait for the next tick. Those fire as triggers, then the loop resumes. A blocked issue, a shipped release, or a meeting that is starting can take the screen now, and hand it back after.

Yes. Pause the agent, change its rules or sources, point it at different screens, or publish over it yourself. An agent does not lock you out of the wall. If the installing user’s access is revoked, their agents pause and surface status rather than keeping publishing.

Yes. Templates are Fugo-authored agent presets with rules and skills pre-filled. You instantiate one by granting data access and picking screens: same Agent entity underneath, not a different product type.

Ask Fugo and MCP help you query and act on your account from chat or external tools. Screen Agents are standing workers that keep running against your screens after you configure them: complementary automation on top of the same Fugo / MCP surface.