UiPath Documentation
test-cloud
latest
false
Test Cloud admin guide

Restricting LLM calls to your own models

Configure the AI Trust Layer to route all LLM calls exclusively through your own models and block UiPath-managed LLMs via Automation Ops policies.

Use this guide to configure the AI Trust Layer so that all LLM calls route exclusively through your own models and UiPath-managed LLMs are never used.

Prerequisites

  • Organization admin access to Admin > AI Trust Layer and Automation Ops.
  • An Integration Service connection to your LLM provider, or credentials ready to create one.

Step 1: Register your models via LLM configurations

Register your models as the authorized LLMs for the tenant at Admin > AI Trust Layer > LLM configurations.

For full setup instructions, refer to Configuring LLMs.

Step 2: Disable UiPath-managed LLMs via an Automation Ops policy

The bring-your-own-model configuration makes your models available but does not block UiPath-managed models. Disable them explicitly via an Automation Ops policy.

  1. Go to Automation Ops > Governance and select or create an AI Trust Layer policy.
  2. In the Models tab, disable the UiPath-managed LLMs you want to block.
  3. Deploy the policy at the tenant level. Tenant-level deployment prevents individual users or groups from overriding the restriction.

For policy configuration details, refer to Settings for AI Trust Layer Policies.

Step 3: Validate enforcement in the audit log

After deploying the policy, verify that no calls are reaching UiPath-managed LLMs.

  1. Go to Admin > AI Trust Layer > Audit.
  2. Filter by Status = Blocked to confirm the policy is intercepting calls intended for UiPath-managed models.
  3. Filter by Model Used and verify that only your configured model appears for calls with a Successful status.

Any call showing a UiPath-managed model name with Status = Blocked confirms the policy is active.

For details on available audit fields and export options, refer to Viewing audit logs.

Step 4: Control fallback behavior

By default, some UiPath products fall back to a UiPath-managed model if your configured LLM is unreachable. To prevent fallback, use one of the following approaches:

  • Policy-based block (recommended): The Models policy you deployed in Step 2 also prevents fallback. If your configured LLM is unreachable and the policy blocks UiPath-managed models, the call fails rather than rerouting.
  • Environment isolation: Place bring-your-own-LLM automations in a dedicated tenant or folder. Use a promotion mechanism so that only approved automations run in the tenant where your model is configured.

Result

After completing these steps:

  • All LLM calls in the tenant route through your configured model.
  • UiPath-managed models are blocked by a policy and cannot be used as a fallback.
  • Blocked calls appear in the audit log with Status = Blocked, providing a verifiable compliance record for security and audit requirements.

Was this page helpful?

Connect

Need help? Support

Want to learn? UiPath Academy

Have questions? UiPath Forum

Stay updated