What's new in IBM® watsonx Orchestrate®

Direct search integration support for Milvus

You can now directly configure your search integration with Milvus. Milvus is a high-performance, highly scalable vector database that you can use for Conversational Search.

For more information, see Setting up search integration with Milvus.

Adding AI assistants from external AI agents

You can add purpose-built AI assistants from AI agents built on third-party platforms to help your users to resolve specific use-cases through the chat. The external AI agent must comply with the API contract of IBM watsonx Orchestrate.

See more details in Adding AI assistants from external AI agents.

Configuring conditional branching in skill flows

You can now configure conditional branches in skill flows to define different execution paths dependent on determined logical conditions, such as variable values. You can also use conditional branches to ensure that your skill flow runs even when a skill fails.

See more information in Set up conditional branches and Continue on error.

More apps and skills

This release adds new skills to the skill catalog. See the list of new skills per app:

• Zoho Recruit

  • Skills
    • Associate candidates with job openings
    • Create a candidate
    • Create a job opening
    • Get all applications
    • Get all applications by record ID
    • Get all candidates
    • Get all candidates by advanced filters
    • Get all departments
    • Get all job openings
    • Get all job openings by advanced filters
    • Get all users
    • Update a candidate status
    • Update a job opening status
      
      

Other enhancements and changes

The following table details the enhancements that are delivered in this release and the user role that is affected by the enhancement.

The following table details the fixes that are delivered in this release and the user role that is affected by the fix:

Documentation updates

See the updates in the watsonx Orchestrate documentation and other contents for this release.

• As IBM Software Hub is called IBM Software Hub from the version 5.0.3, all the labels in the watsonx Orchestrate documentation were updated to reflect this change.

New sections

• The Testing skills and skill flows section shows how to test skills and skill flows to ensure functionality and check if refactoring efforts are needed.

Updated sections

• The tutorial Discovering an RPA bot as a custom skill now has a note about watsonx Orchestrate not suppoting IdP-routed authentication from IBM RPA.

• The topic Feature parity for IBM watsonx Orchestrate is now updated with the new features delivered as part of the IBM Software Hub 5.1 release: support for Task list and Skill studio.

• The Previous releases of IBM® watsonx Orchestrate® was rearranged to show old release notes per year.

• The Managing skills in the AI agent chat section has new instructions to request access to the Orchestrate Agent Skillset.

• The Uploading multiple files content was updated with a new JSON example to show how to configure the input to upload multiple files.

• The Known issues and limitations topic now has the following entries:

  • AI agent chat

    • Adding skills directly to the chat through the Orchestrate Agent Skillset
      
      

  • Prebuilt apps and skills

    • Configuring prebuilt skills for the AI agent
      
      

  • Skill-based actions

    • Skill execution fails with an internal server error for OAuth2 authentication
      
      

Introducing dynamic background

The watsonx Orchestrate platform received a dynamic background on the home page for admins and builders. The new design is based on the IBM Carbon UI standards and is more intuitive for admins to change the chat setting from legacy chat to AI agent chat for the team. Look at the new background!

Enhancements and changes in Skill studio

In this release, the following additions and changes were made to Skill studio:

Using document extractors in Skill studio (Beta version)

You can now create document extractors to define and extract specific fields from rich document formats such as PDFs, Microsoft Word documents, and images. Powered by a large language model (LLM), the configured document extractor identifies and extracts instances of the fields and highlights these fields directly within the document for review.

See more information in Creating document extractors.

Evaluating the performance of your AI assistants

You can evaluate and analyze the performance of your AI assistant by uploading a comprehensive and relevant collection of utterances, in addition to send the utterances to your AI assistant in a test run.

For more information, see Evaluating the assistant.

Adding skills directly to the AI agent

Admins can now add apps and their skills to the AI agent chat to directly use the capabilities of the added skills in the chat. When an admin adds skills to the AI agent chat, they need to provide a description with the purpose and capabilities of the skill for the AI model to route to the right skill based on the user prompt.

For more information, see Adding skills to the AI agent chat.

Hiding the AI assistants from the AI agent chat

From this release, admins can hide the AI assistants from the AI agent chat for the user. This configuration is available for AI assistants that were added to the AI agent chat.

See more information in Hiding AI assistants from the chat.

Registering AI Agents from third-part platforms with the standardized API contract

Register AI agents that come from external third-party platforms and that adheres to the watsonx Orchestrate API contract to the AI agent chat.

For more information, see the API operations available in the External AI agents method.

Other enhancements and changes

The following table details the enhancements that are delivered in this release and the user role that is affected by the enhancement.

The following table details the fixes that are delivered in this release and the user role that is affected by the fix:

Documentation updates

See the updates in the watsonx Orchestrate documentation and other contents for this release.

Updated sections

• The Adding AI assistants to the AI agent chat received additional information about requirements to add AI assistants to the AI agent chat.

• The Using skills in the AI agent chat section was update to show how to use skills directly in the AI agent chat.

• The Events for skill activities section was updated with a new event that is generated when you export a skill.

• The Getting started with the AI agent chat section was updated to address the process of adding skills to an AI agent.

• The overview of the Getting started has a new content about the Orchestrate Agent skill set.

• The Modeling data section has a new content about Name requirements for data types, attributes, and values.

• The Known issues and limitations topic adds entries for the following sections:

  • Building projects

    • Creating a workflow with an assignment activity that uses expressions
    • Using the Share changes feature in projects with document extractor skills
    • Publishing a ruleflow with functions
      
      

  • Prebuilt apps and skills

    • Using IBM Document processing app (Updated size limit)