Automating SRE Operations with Multi-Agent AI
summary
Platform team builds "Infra Assistant" that answers Slack support queries and executes ops via APIs using a vector DB fed nightly by a cronjob from Slack, Confluence and Bitbucket chunks; it uses retrieval augmented generation plus a multi-agent supervisor/subagent architecture where onboarding, Atlassian and AWS CLI (read‑only) agents run tools and return actions. It serves a 300+ person org, grades answers on a 0–5 scale, sometimes takes minutes to respond, and caught a security‑group networking fault during troubleshooting.
AI-written summary · about
chapters (8)
- 0:00Introduction and Speaker Background
- 1:13Company Overview and Digital Experience
- 1:57Evolution from SRU to Platform Engineering
- 5:26Introduction to Infra Assistant
- 7:06Building and Implementing Infra Assistant
- 10:02Advanced Features and Use Cases
- 21:44Challenges and Future Improvements
- 25:47Q&A Session
related talks
44mFrom 4 Hours to 8 Minutes with AI Agents That Transform SRE...
41mWhat Happens After the POC: Building Agentic Systems That Last
24mAI SRE and Agent Handoffs: incident.io's Collaboration Model
33mStrategic AI Agents for K8s SRE: Beyond Ad-hoc Prompting
36mSRE'ing the SRE AI Agents
17mFrom Silos to Synergy: How Platform Engineering Bridges Teams with AI
26mAI & SRE: Embracing the Ultimate Frontier of Automation
38mBuilding AI Agents and Assistants for DevOps Workflows