AI SRE goes mainstream as doubled deploys flood on-call with pages
The Pragmatic Engineer · rss · 2026-10-28
The Pragmatic Engineer takes a deep dive into the heating-up "AI SRE" space. Context: AI coding agents have doubled PR/deploy frequency at most companies, and Rootly data shows after-hours pages nearly doubled — incident rate rises with change rate, driving demand for better on-call tooling.
Interviews with Rootly, incident.io, and ResolveAI reveal:
- AI SRE mostly accelerates incident investigation: orchestrator agents fan out subagents across alerts, dashboards, runbooks, and logs inside Slack/Teams, while engineers keep decision authority; whether agents should ship code changes remains debated.
- Accuracy: verifying hypotheses with adversarial agents, mixing models, evals and backtesting — but measuring accuracy is company- and even team-specific.
- Speed & cost tradeoffs: model routing, lean prompts, tracking cache hit rate.
- Full autonomy is still stage demos: Microsoft's Azure SRE Agent never materialized beyond BUILD; OpenAI built an internal Sevbot, ING is building in-house tooling; agent triage and page silencing is the real current use.
- For on-call engineers: a new category of AI-related outages is emerging; pages will shrink but get harder, with a deskilling risk — incident-response simulations are recommended.
More from coding & agent
- Ben Horowitz: Databricks' agent system of record lifts accuracy from 50% to 84.5% — irinarish · 2026-10-28
- After auditing his own C code with AI, dev no longer trusts unaudited C/C++ — burny_tech · 2026-10-28
- Snowflake Doubles Down on Agentic RL: Overfit First, Then Scale to 65.7 on SWE-bench — StasBekman · 2026-10-28
- Hook-level agent security wins: Microsoft, Anthropic and OpenAI all shifted from gateways to runtime hooks — jacob_posel · 2026-10-28
- Parsing Claude's Raw SSE Stream into a Forensic Timeline of Agent Tool Calls and Diffs — Additional-Ladder-24 · 2026-10-28
- Supabase launches Compute in private alpha: run AI agents and long-running jobs next to your Postgres data — dshukertjr · 2026-10-28