What Open WebUI documents for enterprise deployments
Open WebUI is an open-source, self-hosted chat front end. It puts one or more model endpoints behind a browser UI a whole team can share, and the organization that deploys it runs and secures it. Its own documentation makes an enterprise case built on three things: identity integration, role-based access control, and audit-ready logging.
Identity comes first. The authentication and access docs describe single sign-on through OAuth 2.0 providers such as Google, Microsoft Entra, Okta and Keycloak, authentication against LDAP or Active Directory, optional two-factor sign-in, and SCIM 2.0 for automated account provisioning and deprovisioning from an identity provider. Authorization is additive by design: an account starts with the permissions of its role, which is Admin, User or Pending, and every group membership only ever adds capabilities. Models, knowledge bases, tools and skills each carry their own access control and stay private until the creator grants access to named accounts, groups or everyone.
Whole-instance access is documented in the sharing guide, which walks through network exposure, a pending-approval queue for new accounts, group creation and per-model grants, and a team-access check that asks an administrator to sign in as a regular account instead of trusting their own view.
Logging rewards a careful reading. The security page states that the request audit log is off until an administrator enables it, and that because the software is self-hosted, the deploying organization owns the environment, the configuration and the compliance obligations. Certifications apply to a deployment, not to the software.
# Open WebUI: the request audit log is off by defaultAUDIT_LOG_LEVEL=NONE # no request audit log is writtenENABLE_AUDIT_STDOUT=true # when enabled, also stream events to stdoutWhere a chat deployment stops and a governed platform starts
Open WebUI is a front end over models. It answers a question, searches the knowledge it was granted, and holds a conversation. For many teams that is the whole job.
It is not the layer where work runs. A chat deployment has no per-session machine where an agent executes a task, no permissions model for the agent itself, and no gate between a proposed change and the system it changes. When an IT or security lead has to put AI in front of a review, those are the questions that decide the outcome: what an agent may touch, what it did, and who approves the result before it lands.
That layer is where Kortix sits. Kortix is the open-source AI Operating System: your agents, their skills, your company memory, and every connector in one git repo you own, with the agents working on real cloud computers.
What Kortix gives a company that has to govern the work
Access in Kortix covers agents as well as people. Per-resource permissions for people and agents. Roles, groups, and an audit trail. For identity, Kortix supports SAML 2.0 single sign-on and SCIM 2.0. Every action is recorded. Reading, exporting and streaming the audit log depend on the plan.
Isolation is per session. One isolated sandbox per session. Each session has its own isolated machine and branch, so two jobs never share a filesystem or a working tree. Kortix on GitHub holds the source you can read, fork and audit.
Changes do not land themselves. Session work reaches main through a change request. Merge is default-deny for agents, so a person reads the diff first. You set the connector rules: each tool call is Allow, Ask or Block, and an Ask holds the call until a person approves it and the agent resumes.
Where it runs is your call. Run it on Kortix Cloud, in your VPC, or on your own on-prem network. Self-host is free, and the self-hosting page covers the setup. Compliance is stated plainly: SOC 2 Type I is held; SOC 2 Type II is in progress. Kortix makes no ISO or HIPAA claim.