Launching today
knoxops
Your AI SRE team for production operations
21 followers
Your AI SRE team for production operations
21 followers
Knox brings AI agents and human engineers into one shared room to crush production incidents together. It reasons over your living infra graph (no stale runbooks) and executes fixes safely through a strict Agent Harness that requires your explicit approval for any high-risk actions. It seamlessly plugs into your base infra (like AWS and Datadog) with zero disruption to your existing tools and workflows.
Hey Product Hunt! 👋
I'm Paul, the builder behind Knox. We spent months on one question: how do you make AI agents trustworthy enough to touch production infrastructure?
The answer wasn't smarter models — it was the engineering around them.
Here is how Knox is different:
It actually understands your infra: Instead of relying on stale docs, it reads your configs and builds a living semantic graph. It knows how your services talk to each other.
Zero disruption: It plugs into your existing base infra (AWS, Datadog, K8s) with one command. No need to change your team's workflow.
Built for paranoid engineers (like me): Giving an AI "write" access to production is scary. So we built a strict Agent Harness. Knox can investigate all it wants, but before executing any high-risk action, it stops and waits for a human to click "Approve".
We want the PH community to be among our first early adopters! You can skip the waitlist entirely by using the code PH2026 when signing up, which also unlocks 10,000 free credits for you to play around with.
Ultimately, our vision is for developers and SREs to be able to run their daily ops entirely through Knox. I truly hope it brings a refreshingly different (and actually useful) AI experience to your day-to-day grind.
I’ll be hanging out in the comments all day.
I'd love to hear your thoughts, answer your questions, or even hear why you think giving AI terminal access is a terrible idea (I'm ready for it!).
→ https://knoxops.app/
@paul_knoxops Congrats KNOX — this is exactly the kind of AI infra product people have been waiting for.
What I like most is the stance behind Knox: not “let the model do everything,” but “build the right guardrails so it can be genuinely useful in production.” The semantic graph + human approval loop feels like the right tradeoff.
If you’re an SRE / platform / infra person, this one is worth a look.
Curious how people here think about the balance between autonomy and control in prod ops.