Plan and prepare for AI Builder in 2023 release wave 1

Important

This content is archived and is not being updated. For the latest documentation, go to AI Builder documentation. For the latest release plans, go to Dynamics 365, Power Platform, and Cloud for Industry release plans.

Important

The 2023 release wave 1 plan covers all new functionalities planned to be delivered to market from April 2023 to September 2023. In this article, you'll find the product overview and what's new and planned for AI Builder.

Overview

AI Builder allows citizen developers to use, customize, and govern AI models across their organization with a no-code configuration engine. By using AI models, makers can build more powerful applications and automate workflows that previously required human intervention. These applications can also reduce errors and save time in the process. Azure AI backs the capabilities in AI Builder, which allows makers to process documents, text, images, and generate predictions from structured data. Customers who have in-house AI models built by data scientists can also import them into AI Builder to make them accessible for their citizen makers in Power Platform.

A key point of focus in this release wave is facilitating model governance and licensing. We're revising our admin interfaces to make it easier to govern how AI Builder credits are used across environments. We'll also enhance model monitoring so that customers can track how AI Builder credits are used across apps and flows and by who. We also give makers the possibility to move trainable models across environments for added flexibility in deploying models across the enterprise, and supporting model versioning so that customers can improve model performance in an iterative approach. We'll also increase the breadth of pre-trained document processing capabilities to allow makers to process a wider range of document types.

Use AI Builder documentation to get detailed information on how to train and deploy AI models in Power Platform and get inspired with customer stories that highlight how AI can be used to accelerate digital transformation.

For official product documentation and training for AI Builder, go to:

RSS Subscription Updates to AI Builder 2023 release wave 1

Investment areas

Investment areas

Governance and administration
Governance and administration are key themes to enable the deployment of AI solutions at scale in the enterprise. ModelOps, or model operations, typically cover the governance and lifecycle management of AI models. Today, administrators can already control who has access to models and monitor the amount of AI Builder credits being consumed across environments. Makers can encapsulate models in Power Platform solutions to deploy them across environments. They can also share maker rights to multiple owners.

Model management capabilities will also include the ability to move trainable models across environments. This will facilitate ALM scenarios for AI models and give makers additional flexibility to retrain and optimize model performance across multiple environments. On the admin front, we'll improve governance interfaces to streamline credit management across environments and give admins added visibility into how AI models are used across environments, apps, and flows.

Automate with intelligent document processing
Intelligent document processing (IDP) capabilities allow makers to capture, transform, and process data from documents (for example, email, text, Word, PDF, or scanned documents). Using technologies like computer vision, optical character recognition (OCR), natural language processing (NLP), and machine/deep learning, the extracted data can be analyzed, categorized, transformed, and exported to external systems in an end-to-end process using Power Automate.

Our existing IDP solution can already understand how to extract data from three types of documents:

  • Structured documents: Predefined documents with a set layout like a loan application or tax form.
  • Unstructured documents: Free-form documents like a memo or contract.
  • Semi-structured documents: Documents that combine fixed and free-form structures.

In 2023 release wave 1, we'll focus on improving out-of-the-box prebuilt functionalities to give makers the ability to process a wider range of document types and improved governance and monitoring capabilities to process documents at scale.

To learn more about the entire set of capabilities being delivered during this release wave, check out the release plan for AI Builder below:

For application administrators

User-impacting features to the user experience enabled automatically
User-impacting features should be reviewed by application administrators. This facilitates release change management and enables successful onboarding of new capabilities released to market. For the complete list, look for all features tagged "Users, automatically" in the release plan.

Features that must be enabled by application administrators
This release wave contains features that must be enabled or configured by administrators, makers, or business analysts to be available for their users. For the complete list, look for all features tagged "Users by admins, makers, or analysts" in the release plan.