How it works

Five steps, from a read-only role to answers from your own AI agent.

1. Connect read-only

   You create one role in your cloud account in one click. It can list and describe resources and read billing and security findings. It cannot read your stored data or change anything, with a few opt-in exceptions listed on the Security page. AWS is the first supported provider.

2. Checks run

   A full scan runs every night. Checks are deterministic, with no AI guessing, and every finding cites the resource and the evidence.

   Checks for urgent exposures every 15 minutes, day and night, and alerts you right away. Everything else waits for the morning briefing.

3. Urgent alerts

   New critical findings from the 15-minute checks go to Slack right away, at any hour. Criticals the nightly scan finds alert as soon as it finishes. Without Slack, the owners get an email. Owners can switch urgent alerts off per connection in settings.

4. Morning briefing and weekly report

   Exceptions only, in Slack and email. Skipped when nothing changed.

5. Ask it anything

   Point your own AI agent at OpsAgent over MCP, or chat with it in the dashboard using your own model key. Connect your agent.