Security
This page states what ServiceOps guarantees. It describes commitments and their observable behavior; it deliberately does not describe internal mechanisms.
Tenant isolation
Section titled “Tenant isolation”Your organization’s data is isolated from every other organization’s on every query, by the platform itself — not by each feature remembering to filter. The client cannot choose its tenant: your identity determines your organization server-side, and a forged header or parameter cannot widen it. This holds across tickets, knowledge, search, AI retrieval, files and the API — an API key, for instance, can never read another organization’s data whatever parameters it sends.
New features inherit this boundary automatically, and it is covered by dedicated automated tests, including for AI retrieval.
Authentication
Section titled “Authentication”- Password sign-in uses modern, salted password hashing; passwords are never stored or logged in clear.
- Sessions use short-lived tokens with rotating refresh, delivered in HTTP-only cookies.
- Single sign-on delegates authentication to your Microsoft or Google tenant, and can be made the only way in for your organization.
Authorization
Section titled “Authorization”Every request is authorized server-side against the caller’s roles and permissions — the interface hiding a button is never the enforcement. Agents’ ticket visibility is additionally scoped to what is theirs to see, and customer-company users are hard-scoped to their own company’s tickets.
Secrets
Section titled “Secrets”Credentials your organization entrusts to ServiceOps — mailbox OAuth secrets, IMAP passwords, AI provider keys, data-source tokens — are encrypted at rest and write-only: once saved, they are never returned to any interface again. Logs never carry them.
AI-specific boundaries
Section titled “AI-specific boundaries”- The AI answers outsiders from public knowledge only; internal articles cannot leak through an AI reply.
- Data source calls execute only against the endpoints you declared — redirects and DNS tricks cannot re-route them into internal infrastructure — and API responses are treated as content, never as instructions to the model.
- Every AI action is attributed and ledgered.
Auditability
Section titled “Auditability”Administrative and sensitive actions are recorded in an audit trail: who, what, when. Ticket history records every status change and actor — human or AI — by name.
Self-hosting
Section titled “Self-hosting”On your own infrastructure you additionally control the network perimeter, TLS, and backups — see Self-hosting. The application-level guarantees above are the same in both deployment models.
Reporting a vulnerability
Section titled “Reporting a vulnerability”Report suspected vulnerabilities to your ServiceOps contact or support address — reports reach the engineering team directly, and we appreciate coordinated disclosure.