[ FAQ ]

Short answers

Safety and control

Does it install anything on my servers?: No. No agent, no service, no open port. HelioStack acts through Azure's own managed Run Command.
Can the AI run whatever it likes?: No. The AI never holds Azure access; it can only ask. Every step is checked against your rules, signed, limited to five minutes and tied to one task, and only the broker beside your servers can carry it out. One step at a time
What if the AI is wrong?: The safety comes from the rules and checks around it. New fixes are proved on an isolated test server and approved by a person. A backup comes first and a health check after; if a fix doesn't work, HelioStack undoes it and the team is told. See a fix fail and get undone
Who approves?: A second person, named by your rules. Two people for bigger production changes. Never the AI. Who approves, and what they see
Can I stop it?: Yes. The emergency stop stops work on one server, a group of servers, or everywhere, at once. Everything waiting is cancelled, and lifting the stop releases nothing. The emergency stop

Does my data leave my tenant?: HelioStack's data lives in your own Azure tenant, and its AI models run in the Azure region you choose. When it researches an unknown error online, it does so from Microsoft's cloud, after removing hostnames, addresses, user names and secrets. Security and privacy

Which servers does it support?: Windows and Linux servers on Azure. AWS: coming later
How long until it fixes things on its own?: Coming laterRunning fixes on its own is a later phase. When it comes, only fixes that have worked many times, on many servers, may run on their own, in the hours you choose. One failure sends a fix back to needing a person. Some things always need a person. Trust is earned
Can I use it from Teams?: Coming laterMicrosoft Teams and other chat tools, through your organisation's own AI assistant, come in a later phase. Today you use HelioStack in the browser, or through its documented API.