A recurring job with a result you can inspect

Scheduled AI agents

Run a tested agent on a schedule by attaching a cron expression, timezone and input variables to its frozen workflow. A morning news report is one example. Review the result and destination first, then enable recurring runs. Keep the schedule tied to a known compilation and monitor failures and usage.

Build this agent Read the recipe

Run AI agent on a schedule

Define the morning report
Choose a bounded topic list and sources. Specify whether the result should be a readable digest or structured JSON for your application. Include a source URL and observation time for each finding.
Test and freeze the workflow
Run the job once with representative inputs. Review the output, actual tool actions and cost. Freeze the accepted execution and test that compilation before enabling repetition.
Set cron, timezone and variables
Use the documented compilation update below. The example runs at 09:00 UTC on weekdays. Set your intended timezone explicitly and provide variables that match the frozen input schema.
Monitor and pause when necessary
Review execution status and usage. Three consecutive failures pause a schedule and notify its owner. Resolve the cause before resuming; clear cron_schedule to remove the schedule. Reconcile an ambiguous scheduling response before repeating the change.

Documented schedule update - source documentation

curl -fsS --max-time 30 -X PATCH -H "X-API-KEY: $FLYMYAI_API_KEY" \
  -H 'Content-Type: application/json' \
  --data '{"cron_schedule":"0 9 * * 1-5","timezone":"UTC","schedule_variables":{}}' \
  "$FLYMYAI_AGENTS_API_ROOT/compilations/$COMPILATION_ID/"
# Clear it through the same route with {"cron_schedule":""}.

Set FLYMYAI_API_KEY, FLYMYAI_AGENTS_API_ROOT ending in /api/v1/agents, and your tested COMPILATION_ID. Empty schedule_variables is valid only if your workflow needs no required variables.

What you get

Illustrative output - example data

report_date: 2026-10-23
items:
  - topic: example competitor
    change: new public announcement
    source_url: https://example.com/news
delivery: configured destination

Where the result goes

Keep the report as the execution result for API consumption, or add an explicit delivery tool to the agent. A schedule does not automatically create an email, Slack or database integration.

Cost and operating limits

Each scheduled execution consumes its normal token, tool and model usage. Start with one small schedule and inspect its run history. A cron expression is not a spending cap; keep the workflow and source list bounded.

Check current pricing

Choose the right approach

Implementation choices for this job
Approach	What to evaluate
FlyMy.AI	A frozen agent plus cron, timezone and variables; keep output delivery inside the configured workflow.
A cron job you operate	A deterministic script can work well when you want to maintain its hosting, retries and credentials.
ChatGPT Tasks, Lindy or Relay	[[FACT NEEDED: current schedule, integration and pricing comparison for ChatGPT Tasks, Lindy and Relay from approved sources.]]

Frequently asked questions

Will my laptop need to stay open?

The configured FlyMy.AI schedule runs in the cloud. Your authoring client is separate from the scheduled workflow.

Can I return JSON to my application?

Define and test the required result shape before freezing. Read the execution through the documented API or SDK; do not assume a schedule creates a delivery endpoint.

How do I stop it?

Clear cron_schedule on the existing compilation to unschedule, or use schedule_paused to pause while retaining the cron configuration.