Govern how
AI is used.

Set policies by organization, team, and user, connect your identity provider, and see activity in real time, with auditable logs and custom alerts.

See Liminal

The confidence that comes with
full visibility and control.

Security teams need to know how AI is being used across the organization, and to prove it when anyone asks. Liminal applies your policies to every interaction and records each one, so you can show who used which model, when, and what was protected.

Policies, permissions,
and a complete record.

Set how each type of sensitive data is treated, control which roles and groups can use which models, and review a full log of every prompt and how it was protected.

Liminal Logs and policies view combining three modules: a log of prompts beside the cleansed version sent to the model, with names and account numbers replaced by placeholders, plus user and model; policy controls setting each data type to redact, mask, or warn; and user groups showing each user's role, groups, and approved models.Liminal Logs and policies view combining three modules: a log of prompts beside the cleansed version sent to the model, with names and account numbers replaced by placeholders, plus user and model; policy controls setting each data type to redact, mask, or warn; and user groups showing each user's role, groups, and approved models.

How it works

  1. 01

    Connect your identity provider

    Liminal inherits the teams, groups, and permissions you already manage, so there's nothing to rebuild. People sign in with the credentials they already use.

  2. 02

    Set roles and policies

    Role-based access control, including custom roles, lets you decide who can use which models and who can manage what. Granular security controls let you decide how each type of sensitive data is treated, down to the team and user.

  3. 03

    Monitor in real time

    Streaming logs and custom alerts show how AI and your data are being used, and flag sensitive data events as they're detected.

  4. 04

    Prove it

    Auditable logs give security, compliance, and audit teams a complete record.

From oversight
to impact.

Governance shows how AI is used. Insights shows what it's worth.

Questions from
technology leaders.

Which identity providers does Liminal support?

Liminal supports single sign-on through SAML 2.0 with any compliant identity provider, including Microsoft Entra ID, Okta, and Ping Identity.

Can we control which models each team uses?

Yes. Each model can be assigned to the whole organization, specific teams, or individuals, and each one can carry its own data protection policies.

Can we keep AI activity logs in our own systems?

Yes. Admins can export prompt and activity logs on demand or on a schedule, so a copy lives in your own storage under your own retention policy. Download logs as CSV, JSON, or XML with keyword and date filters, or schedule automatic exports to AWS S3, Azure Blob Storage, or Google Cloud Storage every 15 minutes, hourly, or daily. Your SIEM or BI tools can pick them up from there.

Can alerts connect to our existing security tools?

Yes. Liminal sends email alerts when chosen types of sensitive data appear in prompts. Scheduled log exports can be ingested into security tools such as a SIEM from your cloud storage.

Know exactly
how AI is used.

Tell us about your team and we'll show you how Liminal fits your environment.

See Liminal