September 2025

September 23, 2025

Improved LLM configurations

You can now configure LLMs at the feature level within a product. Products can support multiple LLMs simultaneously, enabling different features within the same product to use distinct models.

The Bring your own LLM option for Agents now supports more connectors: Amazon Web Services, Google Vertex, and Microsoft Azure OpenAI. Previously, only OpenAI V1 was supported.

This release also introduces some changes in field names in the configuration panel, while keeping the configuration procedure just as simple as before.

Extended availability

The LLM configuration feature is now available on more licensing plans, as follows:

• Unified Pricing: Enterprise Platform, Standard Platform, Basic Platform, .
• Flex: Advanced Platform, Flex Standard Platform.

For details, refer to Configuring LLMs.

September 22, 2025

PII In-Flight Masking is now GA

PII In-Flight Masking is now generally available in the AI Trust Layer AI Governance panel.

This feature enables real-time pseudonymization of personally identifiable information (PII) before prompts are sent to LLMs. Detected entities, such as names, phone numbers, or emails, are replaced with contextual placeholders and automatically restored (rehydrated) in the LLM response. This ensures strong data privacy protections without disrupting Gen AI performance or downstream processes.

PII masking is supported for: UiPath Gen AI activities and Agents that route input through LLMs. Once enabled, you can configure which PII entity types (e.g., SSNs, emails, GPS coordinates) to mask and set confidence thresholds per entity.

PII masking helps you:

• Protect sensitive data before it leaves your environment.
• Maintain downstream output integrity via automatic rehydration.
• Provide full traceability via the AI Trust Layer Audit logs.

For details, refer to PII masking.

Feature availability

PII In-Flight Masking is available for the Standard and Enterprise licensing plans.

During public preview, this feature was accessible to all users. With general availability, the following changes apply:

• Enterprise and Standard tiers: Users can continue to configure and use PII masking as before.
• Non-Enterprise tiers:
  • Existing PII masking policies will remain visible in the AI Trust Layer policy.
  • Any non-PII configurations (like model settings) in the same policy will continue to function.
  • PII masking will no longer be applied. Inputs and outputs will appear unmasked in the Audit logs, regardless of what the policy states.

September 17, 2025

Docs experience UX update

We simplified the docs header (Help menu > Documentation) to align with other services.

• Search and Subscribe are now in the secondary navigation.
• Feedback icon was removed (page-level feedback remains at the bottom of each page; general Automation Cloud feedback is available under the Help menu).
• Docs search still opens the smart search panel, now with its own icon.

This update works towards providing a cleaner, more consistent experience across services.

Bug fixes

SAML provisioning rules for group assignments were not recorded in the audit log.

September 1, 2025

Healing Agent available for self-serve trial

If you are on the Flex Enterprise licensing plan and do not already own the Healing Agent add-on, you can start a free 60-day trial. This will allow you to evaluate how Healing Agent can help decrease your automation total cost of ownership by boosting UI automation resilience and minimizing maintenance efforts.

You will receive 200 heals per month for the full duration of the trial.

To start the trial, go to the Trial services tab on the organization-level Licenses page. For more details, refer to Requesting a service trial.

New label for organization administrators

In the Accounts & local groups experience, you can now see the Org admin label next to users with the organization administrator role.

Bug fix

We fixed a bug in the Account & local groups view that was sorting users by email instead of their name, as intended.

Deprecation timeline

We recommend that you regularly check the deprecation timeline for any updates regarding features that will be deprecated and removed.