Skip to main content
Loops are scheduled agents for recurring work: daily reports, alerts, digests, and reviews, built with natural language. A Loop stores a prompt, a cadence, and the context it needs, then runs automatically on your behalf.

​
Creating a Loop

Ask for work that includes a time reference:
Doe detects the schedule, extracts the recurring work, validates the cadence, and asks you to confirm. After confirmation, the Loop appears on the Loops page. Each run starts from the Loop’s source session, so the agent re-executes the task with fresh data every time instead of replaying an old answer. Loops also remember the intelligence mode that was active when they were created. Loops can use the same context as normal requests, including files, folders, previous sessions, and connected integrations. Use Loops for the work that should happen whether or not someone remembers to ask: pipeline checks, weekly metrics, expense reviews, renewal risk, customer follow-ups, and team updates.

​
Managing Loops

Open the Loops page with Cmd + J (Mac) or Ctrl + J (Windows). Each Loop shows its status, cadence, next run, last run, total runs, success rate, and services involved. Right-click any Loop for quick access to Run Now, Pause/Resume, and Delete. When editing a Loop, Doe validates schedule changes before saving. If the prompt no longer includes a usable schedule, Doe will ask you to rewrite it in plain English.

​
Corrections That Stick

Fixing a mistake inside one run does not change future runs. To make a correction permanent, ask Doe to add it as a Loop constraint:
Constraints are injected into every future run exactly as written, and each Loop can hold up to 20 of them. Doe also builds a requirements checklist from the Loop’s spec and verifies the hard requirements in each run’s output before finishing.

​
Operate Loops at Scale

The Loops page is built for more than a short personal list. Use it to monitor recurring tasks across services, spot failures, and manage many Loops at once. Use health and next-run views to catch overdue or failing Loops before they become silent gaps in reporting.

​
Execution Limits

Loops have built-in safeguards:
  • Minimum interval: 5 minutes between runs
  • Active Loops: 50 active Loops per user
  • Max executions: Optional cap on total runs when configured
  • Expiration date: Optional end date
  • Auto-pause: Stops after 5 consecutive failed or connection-blocked runs
When a Loop hits its limit or expires, it pauses automatically. Doe also pauses Loops proactively when a connected service they rely on expires or is disconnected, so runs are not wasted on work that cannot finish. Resuming a Loop clears its failure streak, and a successful run resets the counter.

​
Examples

Sales Pipeline Review - Weekdays at 9am
Daily Expense Check - Every day at 8am
Customer Follow-Up Queue - Every weekday at 4pm
Weekly Metrics Workbook - Mondays at 7am
Executive Briefing - Mondays at 8am

​
Troubleshooting

Check status on the Loops page. It may be paused, expired, or hit its max runs. Also verify your integrations are still connected in Settings -> Connections.
Loops pause after 5 consecutive failed or connection-blocked runs, and immediately when a service they depend on expires or is disconnected. Check run history for errors. Common causes are expired integrations, missing permissions, insufficient credits, or prompts that need clearer scope. Loops belonging to a member who leaves the organization are paused automatically.
Verify your timezone in the Loop settings. Scheduled times use the timezone you selected, not your system timezone.
Yes. Mention the files, folders, sessions, or integrations the Loop should use. Make sure any connected services stay active.

​
Next Steps

Prompting

Write better prompts for Loops

Integrations

Connect apps for Loops to use

Sharing

Share Loop outputs and recurring report folders

Preview Pane

Monitor Loop execution in real-time