[Preview] Document Understanding and AI Center consumption dashboard

Overview

Based on the actions performed in Document Understanding™ and AI Center, the dashboard offers a detailed view of the AI units consumption at tenant level.

Note: For more information on Insights data retention, check the Data retention policy section.

Note:

• Data is refreshed every 24 hours.
• If the value for Automation Name is displayed as Unknown, it means that there is no related data. This could result from using an older Robot or activity package version which is not reporting the needed data, or if the data resides in a different region. Make sure you update your Robot and activity packages to the latest version, so that the dashboard can collect all required data points.
• The dashboard displays all data from a tenant and from the region the tenant is available in.
• The dashboard includes AI units consumption data for Document Understanding and AI Center.

  For Document Understanding, the data covers both cross-platform DocumentUnderstanding.Activities and IntelligentOCR.Activities, applicable for both classic and modern projects, as well as consumption occurring by using API calls. Data is reported for pre-trained specialized and generative models, as well as custom models.

Prerequisites

To access the Document Understanding and AI Center AI units consumption dashboard, Insights must be enabled on your tenant.

For more information, check the Getting started with Insights dashboards page.

Make sure you are using these minimum versions or higher:

• UiPath Studio and Robot: 2023.6
• UiPath.DocumentUnderstanding.Activities package: 2.9.1
• UiPath.DocumentUnderstanding.ML.Activities package: 1.28.2
• UiPath.IntelligentOCR.Activities package: 6.19.1
• UiPath.MLServices.Activities: 2.0.2

Note: AI units consumed by the AI Center ML skills, based on out of the box packages that are older than version 22.10, are not displayed in the dashboard. Only the AI units consumed for infrastructure are displayed. AI units used for calling skills are not tracked and therefore are not displayed in the dashboard.

Note that upgrading the ML packages version also implies retraining for retrainable models.

Filters

You can choose a specific time period to review AI units usage.

Metrics

Tile	Description
Consumption by project chart	Consumed AI units, grouped by Document Understanding or AI Center project. The information is displayed in the form of a line chart.
Consumption by project table	Consumed AI units, grouped by Document Understanding or AI Center project. The information is displayed in the form of a table with additional information: Project Name: name of the Document Understanding or AI Center project used for consuming AI units. Project Id: ID of the Document Understanding or AI Center project used for consuming AI. Project type: type of the project used for consuming AI units: Document Understanding modern project Document Understanding classic project AI Center N/A - this is displayed for IntelligentOCR workflows that do not reference a Document Understanding project. AI units: total number of AI units consumed in the project. Pages: total number of pages processed in the project.
Consumption by operation	Consumed AI units, grouped by Document Understanding or AI Center project and operation: Project Name: name of the Document Understanding or AI Center project where the operation was run. Project Id: ID of the Document Understanding or AI Center project where the operation was run. Operation name: name of the operation used for consuming AI units (for example, model name, digitization, Intelligent Keyword Classifier, and so on). Automation name: name of the automation used for consuming AI units. Note: If the value is displayed as Unknown, it means that there is no related data. This could result from using an outdated Robot or activity package version, or if API Keys were used to access public endpoints, such as du.uipath.com/invoices. AI units: total number of AI units consumed by the operation. Pages: total number of pages processed by the automation.
Consumption by AI Center operation chart	Consumed AI units corresponding to AI Center operations: Runtime consumption referring to predictions using AI Center models (excluding Document Understanding). Training consumpation for training the model. Hosting for serving the model.
Consumption by AI Center resources table	Consumed AI units corresponding to the used AI Center resources, including the operation performed. The information is displayed in the form of a table with additional information: Project Name: name of the AI Center project where the operation was run. Source: this can be one of the following: Name of the ML Skill used for hosting. Name of the ML package used for training. Name of the ML Skill used for runtime. Type: the type of the operation used for consuming AI units: Runtime Training Hosting AI units: total number of AI units consumed by the resource.