July 2026

July 2026 release notes for Unstructured and Complex Documents in Automation Cloud.

For IXP service-level updates, refer to the IXP release notes.

For other IXP capability updates, refer to the Communications Mining release notes.

July 27, 2026

Unstructured and Complex Documents availability in Singapore

The UiPath® IXP capability, Unstructured and Complex Documents, is generally available in the Singapore region.

For more details, check the following resources:

• Global cloud regions
• AI features and model routing

Bug fixes

• Project scores shown in the annotation screen could differ from those on the Measure and Build pages. Scores are now consistent across these views.

July 23, 2026

To folder deployment option (preview)

You can deploy an IXP model version to an Orchestrator folder using the To folder deployment option. This option is required to use IXP models as solution resources.

IXP models deployed to folders are available to automation developers in the Flow canvas and within the Document Understanding 3.4.0-preview and Intelligent OCR 7.4.0-preview package versions.

For more details, refer to the following resources:

• Publishing model versions
• Document Understanding activities release notes
• Intelligent OCR release notes

IXP models in Solutions (preview)

IXP models are available as solution resources. You can use the models in Flow, and within the Document Understanding Project Extractor and Extract Document Data activities.

The selected IXP model is added to the solution. Then, it is packaged and deployed together with the rest of the solution's resources.

July 21, 2026

Audit of LLM inputs and outputs in AI Trust Layer

The Unstructured and Complex Documents capability now records its document extraction operations in the UiPath AI Trust Layer™ audit log. For more details, refer to:

• Auditing LLM interactions
• Viewing audit logs

Document error rate consistency

The document error rate displayed on the Build page now uses the same calculation as the document error rate on the Measure page. This ensures consistency across both views. Previously, the two pages could show different values for the same model version because the document error rate was calculated differently.

In some cases, you may notice a document error rate greater than 100%. For details, refer to Evaluating model performance.

Bug fixes

• An inflated error rate was displayed when documents had empty field groups with only fields marked as missing. No mismatch warnings appeared in the annotations pane.
• When you annotated a field value whose ground truth differed from the model prediction, the mismatch indicator appeared only after you resubmitted the annotations and predictions were re-run.

July 13, 2026

Bug fixes

• Occasionally, when you used Autopilot to create a project with a large prompt and a large number of documents, the taxonomy was not generated after you selected Continue.
• In certain cases, the model returned empty field groups, which appeared as extractions with empty field values and reappeared even after you dismissed them. Empty field groups returned by the model are now ignored and no longer displayed.

July 8, 2026

Pricing update

Added July 10, 2026: The pricing updates also apply to Unified Pricing. These were released on July 8, but were not included in the release notes at the time.

AI Unit and Platform Unit consumption for Unstructured and Complex Documents extraction is now charged per operation, instead of a flat rate per page. IXP and Document Understanding licenses can be used together, with no additional cost per page.

For more details, refer to the following resources:

• IXP Flex Pricing Plan
• IXP Unified Pricing Plan
• Metering and charging logic (Flex Plan) for Document Understanding
• Metering and charging logic (Unified Pricing) for Document Understanding