Software Engineering Daily

Agentic DevOps at AWS

49 min

Neha Kazwami, leader of Agentic DevOps at AWS, discusses how AI agents are transforming DevOps by automating incident response, root cause analysis, and operational toil. AWS's DevOps Agent uses state-of-the-art models and integrations with tools like Datadog, Splunk, and ServiceNow to help engineers wake up to a root cause instead of an active incident. The episode covers the origin story, dogfooding culture, the importance of determinism in agentic systems, and the future of SRE roles as agents take on more operational work.

Summarized by Podsumo

🎧 Listen 🎙️ Ask about this episode

✨ Key Takeaways

1

AWS's DevOps Agent achieves 94% root cause accuracy in preview, with internal teams seeing 85-95% success after customization.
2

Agentic DevOps is both a catalyst for change (accelerating code production) and a cure (automating incident response and pipeline fixes).
3

Determinism is still crucial in agentic systems; AWS injects safety through read-only permissions and plans to use automated reasoning.
4

MCP integrations are popular, but AWS recommends centralizing common integrations to scale effectively across large organizations.
5

SREs will shift from pattern matching to handling more complex scenarios as agents take over trivial cases.

💬 Notable Quotes

"Agentic DevOps is both a catalyst for change and the cure for change."
"The engineer wakes up to a root cause instead of an active incident."
"If you haven't tried using any of these agentic tools and models yourself, start there. Use them to make your grunt work go away."