Automation Cloud release notes

DeepRAG (Deep Research-Augmented Generation) is available as a new capability within Context Grounding. It enables retrieval-augmented synthesis of information from documents, allowing agents to generate citation-backed answers to complex queries.

DeepRAG is introduced as an additional search strategy when configuring agent contexts. DeepRAG queries incur separate charges from basic and advanced retrieval operations. For details, refer to Context Grounding licensing.

If your identity provider (IdP) does not support automatically consuming updates from UiPath's SAML metadata URL, and you require signed authentication requests or responses, you will not be able to reliably update verification certificates in your IdP. In this case, you can now upload your own verification certificate via API endpoints, and ensure you meet your IT requirements for signed SAML requests or responses. For step-by-step information on uploading a verification certificate, visit Managing UiPath signing certificate for SAML authentication requests.

In the Accounts & local groups section of your organization, in the User accounts tab, downloading the role assignments list now includes user display names as expected. Previously, the CSV file contained blank cells in the display name column.

You can now hide the left rail navigation icon for Assistant from all non-administrator users in your organization. This change affects only the navigation UI. Users can still access the service if they have the direct URL. For more information, visit Hiding services.

We are introducing key updates to Context Grounding licensing to make pricing more transparent and scalable.

To make Context Grounding more accessible, basic search and retrieval operations are now free of charge. This includes any time an index is accessed, whether by an Agent or a GenAI activity. Additionally, basic ingestion of plain text documents continues to be completely free. These changes significantly reduce the cost for customers using Context Grounding for straightforward queries.

To support more advanced scenarios, we are introducing a new pricing model for advanced ingestion. Ingestion of multi-modal documents, such as scanned PDFs, files containing tables, or documents with embedded images, now incur a cost of 2 AI Units (or 0.40 Platform Units) per 15,000 tokens, roughly equivalent to 10 pages.

Agent executions with Context Grounding are now metered separately using Agent Units, rather than AI Units or Platform Units. This change provides greater pricing clarity and better alignment with usage patterns across GenAI Activities, Autopilot, and Agents.

For details, refer to Context Grounding licensing.

Creating a new tenant in an existing organization caused roles from unprovisioned services to appear in the Check access results. These roles were incorrectly shown as assigned to users, even though the related service had not been added to the tenant.

Agent units consumption is now displayed on the Consumables tab of the tenant-level Licensing page. If no agent units are allocated to a tenant, consumption information is not displayed in the list.

For more details, refer to The Consumables tab.

To ensure a smooth transition from the Flex licensing model to Unified Pricing, organization administrators now receive a one-time notification if users are still assigned Flex licenses after the switch.

The notification prompts admins to review and update user license assignments to ensure they reflect the Unified Pricing model and prevent any potential service disruption.

For more details, refer to Licensing events.

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