Agentic DevOps at AWS
Episode
49 min
Read time
2 min
Topics
Career Growth, Productivity, Startups
AI-Generated Summary
Key Takeaways
- ✓Incident Response Automation: AWS DevOps Agent begins investigating before the on-call engineer is paged. When an alarm breaches a threshold and auto-cuts a ticket, the agent simultaneously starts root cause analysis using a pre-built system topology map. Engineers wake up to a completed hypothesis, ranked mitigation steps, and a full audit trail rather than a blank incident screen.
- ✓Customization Drives Accuracy: Internal Amazon teams using DevOps Agent start at an 85% baseline accuracy but reach the high nineties by adding team-specific runbooks and configurations. The lift is low — not heavyweight integration work — but the payoff is significant. Organizations scaling DevOps Agent should centralize common MCP tool integrations rather than leaving each team to configure them independently.
- ✓Determinism via Permission Boundaries: In a probabilistic agentic system, AWS enforces determinism through strict, read-only permission defaults. Expanding agent permissions requires deliberate, intentional user action. AWS also applies automated reasoning — a mathematical modeling technique that compares system state before and after a change — to detect unintended policy drift without relying solely on LLM inference.
- ✓Triage Agent Deduplication: A companion triage agent consolidates duplicate alarm tickets before the DevOps Agent investigates. If five tickets fire for the same root cause, the triage agent collapses them into one, preventing redundant agent task runs. Since pricing is based on active agent task hours, deduplication directly reduces operational costs alongside reducing engineer alert fatigue.
- ✓SRE Role Shift, Not Elimination: As agents handle routine incident pattern matching, SRE engineers will increasingly engage only on cases where agents produce incomplete or failed analyses. AWS internally still requires engineers with SRE experience to build these products. Mid-career DevOps engineers should adopt agentic tools immediately for daily workflows and deliberately move toward higher-complexity problem domains the agents cannot yet resolve.
What It Covers
Neha Goswami, AWS Agentic DevOps lead with 22 years at Amazon, explains how AWS DevOps Agent automates incident response from alarm to root cause analysis, achieving 85–95% accuracy in internal deployments, while discussing determinism, MCP integrations, and the evolving role of SRE engineers.
Key Questions Answered
- •Incident Response Automation: AWS DevOps Agent begins investigating before the on-call engineer is paged. When an alarm breaches a threshold and auto-cuts a ticket, the agent simultaneously starts root cause analysis using a pre-built system topology map. Engineers wake up to a completed hypothesis, ranked mitigation steps, and a full audit trail rather than a blank incident screen.
- •Customization Drives Accuracy: Internal Amazon teams using DevOps Agent start at an 85% baseline accuracy but reach the high nineties by adding team-specific runbooks and configurations. The lift is low — not heavyweight integration work — but the payoff is significant. Organizations scaling DevOps Agent should centralize common MCP tool integrations rather than leaving each team to configure them independently.
- •Determinism via Permission Boundaries: In a probabilistic agentic system, AWS enforces determinism through strict, read-only permission defaults. Expanding agent permissions requires deliberate, intentional user action. AWS also applies automated reasoning — a mathematical modeling technique that compares system state before and after a change — to detect unintended policy drift without relying solely on LLM inference.
- •Triage Agent Deduplication: A companion triage agent consolidates duplicate alarm tickets before the DevOps Agent investigates. If five tickets fire for the same root cause, the triage agent collapses them into one, preventing redundant agent task runs. Since pricing is based on active agent task hours, deduplication directly reduces operational costs alongside reducing engineer alert fatigue.
- •SRE Role Shift, Not Elimination: As agents handle routine incident pattern matching, SRE engineers will increasingly engage only on cases where agents produce incomplete or failed analyses. AWS internally still requires engineers with SRE experience to build these products. Mid-career DevOps engineers should adopt agentic tools immediately for daily workflows and deliberately move toward higher-complexity problem domains the agents cannot yet resolve.
Notable Moment
Goswami revealed that Amazon's large-scale Java version upgrade — handled centrally using what became AWS Transform — saved millions of hours of developer work. The agent produced pull requests, verified builds, and ran tests on behalf of thousands of teams, representing a fundamental mindset shift in how centralized campaigns operate at scale.
Episode Transcript
AI agents have become capable of reasoning across large amounts of data, calling tools, and taking sequences of actions autonomously. These qualities make them well suited to some of the most persistent pain points in DevOps, including the on call engineer woken at 3AM to diagnose an incident, the build failure that takes hours to trace back to a root cause, and the operational toil of modern software delivery. Agentic DevOps is the emerging paradigm that applies these capabilities to the full software lifecycle, with the goal of matching the velocity of modern software delivery with an equally capable operational layer. Neha Goswami leads Agentic DevOps at AWS and has been at Amazon for over twenty years. In this episode, she joins Matt Merrill to discuss the AWS approach to Agentic DevOps, how Amazon dog foods its own DevOps tooling, how their DevOps agent works from alarm to root cause, why determinism still matters in an Agentic world, creative MCP integrations customers are building, and what the future holds for SRE engineers as agents take on more of the operational work. Matt Merrill is a software engineering leader with over twenty years of experience building and scaling software teams across enterprise and product focused organizations. His background is in back end development, cloud architecture, and distributed systems design. He currently architects and delivers software products and leads a team of engineers at Dept Agency. You can learn more about his work at code. The other mattm.com. Alright. Hey, everybody. I am Matt Merrill, and I'm here today with Neha Goswami from Amazon Web Services. And before we dive into our topic today, which is agentic DevOps, Neha, can you tell us a little bit about yourself and what you do at Amazon? Yeah. Absolutely. Glad to be here, Matt. Let me start with what I do here at Amazon. I lead agentic DevOps in agentic AI organization in AWS, where my team is building products, which bring agents into DevOps and application security aspects of development. I've been in, Amazon for almost twenty two years at this point. And fun fact, I started my journey as an SDE, stayed out of college, in a team called builder tools. So I was building tools, which was mainly focused on internal Amazon developers. And I was there in the team for three years and then moved around in different parts of the company and came back to building developer tools four years ago. And that has been, like, the most fascinating journey for me because I came back. Awesome. Yep. Came back to build the same tool that I was building, but for external developers and in the world of AI. And it's just fascinating to see how this world has changed in all these years. Yeah. Honestly, I've been really looking forward to this. I can't wait to talk about it. And what a great prep for dealing with outside developers. How many developers and engineers does Amazon have roughly? Because, …
Get the full transcript (8,760 words) + summary by email — free
One-time email with the complete transcript and AI summary of this episode. No account needed.
One email, no spam. We’ll also show you what SignalCast does.
You just read a 3-minute summary of a 46-minute episode.
Get Software Engineering Daily summarized like this every Monday — plus up to 2 more podcasts, free.
Pick Your Podcasts — FreeKeep Reading
More from Software Engineering Daily
TypeScript 7 and What Comes Next
Aug 27 · 57 min
In Good Company with Nicolai Tangen
CPP Investments CEO: The Canadian Model, Public vs Private and Investing for 22 Million Canadians
Jul 8
More from Software Engineering Daily
The Gap Between AI Spending and AI Value
Aug 25 · 54 min
Practical AI
AIUC-1: Building trust in AI agents
Jun 25
Books, tools, and gear mentioned in this episode
SignalCast may earn commission on purchases via these links.
Tools
by AWS
“Neha Goswami, AWS Agentic DevOps lead with 22 years at Amazon, explains how AWS DevOps Agent automates incident response from alarm to root cause analysis, achieving 85–95% accuracy in internal deployments.”
by AWS
“Amazon's large-scale Java version upgrade — handled centrally using what became AWS Transform — saved millions of hours of developer work.”
More from Software Engineering Daily
We summarize every new episode. Want them in your inbox?
TypeScript 7 and What Comes Next
The Gap Between AI Spending and AI Value
AI and the New Global Security Landscape
How LLMs Are Reshaping Recommendation Systems
Rebuilding the Cloud for AI Agent Code
Similar Episodes
Related episodes from other podcasts
In Good Company with Nicolai Tangen
Jul 8
CPP Investments CEO: The Canadian Model, Public vs Private and Investing for 22 Million Canadians
Practical AI
Jun 25
AIUC-1: Building trust in AI agents
Machine Learning Street Talk
Jun 22
He won a Nobel here for AlphaFold. Then he left. - John Jumper
Eye on AI
May 15
Why Agentic-First Startups Won't Disrupt Enterprises as Fast as Everyone Thinks | Kris Lovejoy
In Good Company with Nicolai Tangen
May 6
IBM CEO: Transforming a Tech Giant, AI Bets and Quantum Computing
Explore Related Topics
This podcast is featured in Best Cybersecurity Podcasts (2026) — ranked and reviewed with AI summaries.
Read this week's Startups & Product Podcast Insights — cross-podcast analysis updated weekly.
You're clearly into Software Engineering Daily.
Every Monday, we deliver AI summaries of the latest episodes from Software Engineering Daily and 192+ other podcasts. Free for one show.
Start My Monday DigestNo credit card · Unsubscribe anytime