Changelog #17: Wake up your AI agent before getting paged

Every second matters when an incident happens. AI SRE investigates your incidents, pulls all the evidence, and gives you a root cause analysis. Focus on decision making and the fix itself, and leave the data analysis to AI SRE.

Read on to make the most of AI SRE incident analysis.

Wake up to a root cause, not an unknown incident

Paged at 3am with nothing but an alert name? AI SRE starts investigating the moment the incident happens. It checks what's failing and what changed around the start time, then posts a short summary to the incident timeline and Slack. By the time you've opened your laptop, the root cause analysis is ready.

Enable AI SRE investigation for all incidents.

Tell AI SRE how to investigate

Every team debugs differently. Write your own prompt to control what AI SRE checks, how deep it goes, and how it reports back.

Want different prompts for different incidents? Set up multiple AI SRE integrations, give each its own prompt, and trigger the integrations in your escalation policies.

Customize your AI SRE prompts and use them in escalation policies.

Give AI SRE the context it needs

AI SRE builds its root cause analysis from your logs, traces, metrics, errors, and recent releases. Connect GitHub, GitLab, Sentry, Grafana, Datadog, Linear, and Notion, and let AI SRE dig through them during the investigation.

Connect your existing tools to AI SRE.

Not using AI SRE yet?

Investigate your first incident. Trigger an investigation with a single click for any of your incidents.

Want to analyze your data even before an incident is opened? Start a new AI SRE chat and ask about your telemetry, errors, recent commits and changes.

Start using AI SRE.

---

Fresh community content

GPT-5.6 Sol and Ultra Mode: What You Need to Know

OpenAI’s GPT-5.6 ships a three-model family and Ultra Mode, which lets the flagship Sol model spawn and coordinate specialized sub-agents internally. What used to take external orchestration frameworks now happens inside one API call.

GitHub Agentic Workflows: Automating DevOps with AI-Powered GitHub Actions

Write a workflow in plain Markdown and let an AI agent run it as a GitHub Action: performance audits, issue triage, code cleanups. A practical look at continuous AI in the pipeline you already have.

How They Fixed AI’s Biggest Local Hardware Problem (with FreeToken)

FreeToken is an open-source inference engine that runs Mixture-of-Experts models dramatically faster than Ollama once they outgrow your GPU’s VRAM. A look under the hood at expert caching and bandwidth-adaptive loading.

I replaced my entire tech stack with Postgres

Caching, vector search, full-text search, geospatial, cron, and document storage, all inside the one database you already run. Not because you should, but because you can.