← All Webinars

Upcoming Webinar

Close the Loop: Turn Production Failures Into Pull Requests

Your software factory stops at deploy. Watch a failed production check turn into a draft pull request, live.

Date

Nov

10

2026

Time

6 PM CET

12 PM EST

9 AM PST

Add to calendar: Google · ICS

Can't make it live? Register anyway — every registrant gets the recording by email.

Your agents write the code, the tests and the pull request. Each step starts the next one: a ticket starts the agent, a pull request starts the review, a merge starts the deploy. Then the code ships, and the loop stops. Logs and traces only show what your system saw. They can't tell you a customer just tried to pay and nothing happened. Usually a customer tells you, hours later.

Stefan Judis builds the missing step live. A check catches a real failure in production. The signal goes to a coding agent with the evidence attached. The agent follows your runbook, opens a draft pull request, and the same check confirms the fix.

What we'll cover

  1. The production context your agents are missing, and why tests and logs miss what breaks for users
  2. What makes a signal safe to automate on, and what arrives with a failed check
  3. Live: from a broken checkout to a draft pull request, with a webhook, a Claude Code routine and the Checkly CLI
  4. The runbook that tells your agent when to open a PR, when to fix the check, and when to do nothing
  5. How much of the loop to hand to agents, from a Slack message to auto-rollback

What you'll walk away with

  • —A working setup you can copy: webhook, routine and runbook
  • —A runbook pattern that keeps your agent from chasing false alarms
  • —A clear way to decide how far to trust the loop

Speaker

Stefan Judis

Developer Relations · Checkly