Question 1

Does NOCless change anything in my systems?

Accepted Answer

No. Every integration is strictly read-only — its own service account with read permissions, and no write path to a monitored system anywhere in the codebase. What NOCless sends are the messages your operator — or an autonomy policy you set — approved, issues that track open incidents (if you connect a GitHub repository), and its own audit archive to a bucket you choose. Changing systems is a design boundary, not a setting.

Question 2

What happens when the model is wrong?

Accepted Answer

By default no message reaches a person without operator approval: NOCless starts at autonomy level L0, where the agent proposes and your operator approves. Every decision and message is also written to an append-only audit log, so a wrong proposal is reviewable evidence — not an outage.

Question 3

Can it run air-gapped?

Accepted Answer

Yes, on Kubernetes. The offline bundle carries the images, Postgres, the Helm chart and checksums for a cluster without internet access — k3s, AKS, EKS, GKE or your own. NOCless runs with a local model (Ollama, vLLM, LM Studio) instead of a cloud API, makes no outbound calls of its own, and checks its licence offline. No data has to leave your network.

Question 4

Can you run it for us?

Accepted Answer

Yes. As a service, every customer gets a dedicated NOCless instance — its own database and audit log, nothing shared with anyone else — hosted in Amsterdam (EU), so your data is stored in the EU. If you choose a cloud model, its prompts go to that provider; the subprocessors are listed in the data processing agreement. It reaches your systems the same way it does on-prem — read-only, through the service accounts you give it — over Tailscale: the instance joins your own tailnet, your ACL decides which endpoints it may reach, and you can remove it at any time. We take care of upgrades, backups and keeping it running; your operators use the console as they would on-prem.

Question 5

Where does my data go?

Accepted Answer

Alerts, logs and diagnosis context stay inside your deployment — on your infrastructure, or in your dedicated instance in Amsterdam; state and the audit log live in its PostgreSQL. What leaves is what you configure: approved messages to your channels; GitHub issues for open incidents (redacted), if you connect a repository; past audit months to your own bucket, if you set up an archive; and — only with a cloud model — the prompts to that provider. With a local model, even prompts never cross your perimeter.

Question 6

What is NOCless not?

Accepted Answer

It is not an auto-remediation bot. It will not restart your pods, roll back a deploy or "fix" anything — no such action exists in the product. A human always makes the fix decision; NOCless hands them the diagnosis.

Question 7

What does it cost?

Accepted Answer

NOCless is priced per monitored Kubernetes or OpenShift cluster, never per user — DEV, QA and PROD clusters count the same, and support and updates are included. Outside Kubernetes, the offer says what one environment is. Before you commit, a pilot runs for one month at no cost, starting at L0 in shadow mode. Book a demo and bring your hardest shift.

Imprint

About this site