Resources. Docs, guides and comparisons for running production with an AI SRE.
Documentation
Get up and running.
Get started
Sign up, connect your first environment and read your first incident.
Use RubixKube
Incidents, RCAs, Tasks, and how much Rubix may do on its own.
Connect your stack
Clusters, clouds, machines and the tools you already use.
Learn
The ideas behind it.
MTTU (Mean Time to Understand): Definition, Benchmarks, and How to Measure ItMTTU measures how long it takes a team to understand what broke and why. Learn how it differs from MTTR, why it is the real bottleneck in incident response, and how to track it.
What Is an AI SRE?An AI SRE watches your systems, investigates what breaks, finds the root cause and proposes the fix. Here is what it does, how it differs from AIOps, and how to pick one.How it works
Compare
Compare AI SRE tools.
- AlternativesClericCleric Alternatives in 2026The best Cleric alternatives in 2026, ranked: what each one does, how you get it, what it costs, and who it suits.
- AlternativesResolve AIResolve AI Alternatives in 2026The best Resolve AI alternatives in 2026, ranked: what each one does, how you get it, what it costs, and who it suits.
Use cases
What teams use it for.
- Investigate a failing podOOMKilled, CrashLoopBackOff, ImagePullBackOff: ask Rubix what happened and get the cause with the evidence.
- One incident for an alert stormAlerts from Datadog, PagerDuty and Grafana are triaged and grouped, so one cause becomes one incident.
- A root cause you can checkEvery RCA lays out the cause step by step and cites the logs, events, deploys and pull requests behind it.
- Fix it with a pull requestRubix proposes the fix as a pull request or a change through your tools, and waits for your approval.
Blog
From the blog.
- Nov 4, 2026
What needs you: rebuilding RubixKubeEvery team that runs infrastructure has a few people who hold the whole system in their heads. AI changes who carries that map. So we rebuilt RubixKube around the two questions that matter most: what needs me, and what can I do about it? - Sep 7, 2026
The Drill and the Map: Why AI for Operations Can't Live in One RepoA coding agent knows one repository intimately. An operator has never had the luxury of one repository. The knowledge problem in software operations is breadth across codebases and continuity over time, not depth in a single directory. Every AI developer tool on the market today is a drill. The operator needs a map. - Sep 4, 2026
The Last Unserved User: Why Nobody Ever Built a Tool for the OperatorEvery tool in operations was built for the infrastructure or for the organisation. Not one was built for the person holding the pager.
The operator is the last unserved user in engineering.