Aptible’s Post

Aptible reposted this

View profile for Henry Hund, graphic

Building AI SRE Agents to fix on call and incident response

There’s a game-changer for production debugging: AI. So many engineering teams are struggling to launch an AI Agent for SREs, so we wrote a guide on how to build one 👇 Since this summer, we’ve been building our own AI Agent internally to help our team fix bugs, investigate incidents, and resolve production issues faster (and without pulling in other team members or more senior engineers). We did this by creating an Agent that doesn’t try to answer your questions for you or tell you what to do next. Instead, it surfaces all the relevant information that you need to resolve the problem yourself. This has saved our team HOURS of time digging through logs, metrics, docs, recent deployments, etc. And because it’s been so helpful for us, we want to share how we did it 😄 We’ve put together a 3-part guide that shows you how to build your own AI Agent to help production engineers debug and fix production issues as quickly and efficiently as possible. You can find the link to parts 1 and 2 in the comments below 📌 Part one shows you how to: 1. Set up a new application in Chainlit 2. Connect your application to an LLM Part two is how to: 1. Make your Agent faster and add real-time “typing” (similar to ChatGPT) 2. Give your Agent a personality and specialization Stay tuned for part 3 next week! If you want a sneak peek at it, connect with me and send me a DM and I’ll share the link to the private page.

  • No alternative text description for this image
Like
Reply
Emil Peykov

Senior DevOps | Linux System Administration, Network Administration

1mo

So you build your AI first , then you fire your SREs 😅

Like
Reply
See more comments

To view or add a comment, sign in

Explore topics