This article covers the prerequisites for Conversational AI, powered by Matrix42 Intelligence, across Matrix42 Enterprise, as well as instructions for configuring each component.

Prerequisites

This article describes everything you need to set up Matrix42 Intelligence properly.

1. Verify that you have the following product version and extensions:
   1. Matrix42 Enterprise. Specifically, you require version 26.1.0.3224. This is a minor release shipped on July 23, 2026.
   2. AI Intelligence Foundation (extension)
   3. Agentic Enterprise Assistant (extension)
   4. AI Assist for Service Desk (extension)
2. Configure the Matrix42 Intelligence connection using your organization's credentials:
   1. Subscription ID
   2. Client ID
   3. Client Secret
   4. User Client ID
   5. User Client Secret
   6. Update Private Key
3. Activate and validate the new experience:
   1. Verify the Matrix42 Intelligence service connection.
   2. Enable the Matrix42 Intelligence AI Service Provider.
   3. Configure access for UI Actions for the relevant roles and user groups.
   4. Review where the functions are displayed.
   5. Review any customized ticket previews or layouts.
   6. Test the functionality with a designated Service Desk user.
   7. Release the functionality to the intended users. 

Conversational AI Configuration

Configure the Matrix42 Enterprise Connection

Before any component can be enabled, you need to connect Matrix42 Enterprise to Matrix42 Intelligence through a Matrix42 Intelligence AI Service Provider and a Service Connection.

1. Install the Matrix42 Intelligence Chat Extension from the Extension Gallery.
2. Enable the Matrix42 Intelligence AI Service Provider. No custom development is required.
3. Create and configure the Service Connection, including the endpoint and required connection parameters, through the supported administrative mechanisms.

Conversational AI only becomes visible where the AI Service Provider is enabled, the Service Connection is correctly configured, and the relevant capability is assigned. If any of these is missing, the conversational entry point simply does not appear, rather than showing a broken or partially loaded state.

Capability Activation

Conversational AI is activated per capability. Two capabilities are available, and where each one shows up depends on which capability is assigned:

1. Open the capability configuration for Conversational AI in the platform configuration model.
2. Assign the relevant capability (or both) depending on which experience should be enabled.
3. Save and verify visibility in the target application.

• firmation interaction patterns in M42 Assist are reused, so agents do not need to learn a new interaction model.

M42 Assist (Agents)

Once the platform connection is configured and the Conversation AI for Agents capability is assigned, Conversational AI appears automatically inside the existing M42 Assist panel, with a clear action for starting a conversation. No separate standalone chat interface is required.

• No separate agent-facing configuration is required beyond capability activation.
• Context is passed automatically: authenticated user, current language (where supported), current object (when opened from an object preview or dialog, where technically available), and current application or view (when opened from a report or list).
• This capability is available in all applications other than the Self Service Portal. Confirm it is assigned for each application where agents should have access.
• Existing AI-result and confirmation interaction patterns in M42 Assist are reused, so agents do not need to learn a new interaction model.

Self Service Portal (End Users)

Enabling the Conversation AI for End Users capability adds a natural-language conversational entry point to the Self Service Portal. Exactly how it appears depends on the portal's Unified Search configuration.

1. Confirm the Service Connection is set and the AI Service Provider is enabled (see Platform Connection above).
2. Assign the Conversation AI for End Users capability.
3. Check the Unified Search setting for the relevant Navigation Item, under Admin > Global Settings (this is a dynamic, per-navigation-item setting).

The display behavior follows this logic:

If Matrix42 Intelligence is installed and configured, default activation for the Self Service Portal follows the agreed product configuration. If the Matrix42 Intelligence service is temporarily unavailable, the portal falls back to Global Search and logs an error; it will not show a broken or permanently loading component, even if the configuration is incomplete or invalid.

Microsoft Teams (End Users)

Microsoft Teams integration requires additional setup beyond capability activation, since it involves connecting a Teams environment and mapping identities between Teams and Matrix42 Enterprise.

1. Enable the Conversational AI capability for Microsoft Teams.
2. As a customer administrator, connect your Microsoft Teams environment to Matrix42 Enterprise, following the prerequisites, permissions, and application registration steps in the Teams setup documentation.
3. Configure user mapping so each Microsoft Teams identity resolves to the corresponding Matrix42 Enterprise user. Requests are then executed in the context of that authenticated user through impersonation, respecting the user's existing Matrix42 permissions and data access.
4. Verify authenticated end users can open the conversational experience from Teams, ask a question, get guidance for a supported simple issue, report an unresolved issue, or request a supported simple service.

A successfully reported issue or submitted request creates the corresponding Matrix42 record, and the assistant confirms this with an available identifier or reference. Unsupported, complex service-request scenarios do not create incomplete records.

Workflow Designer (Workflow Administrators)

Conversational capabilities in Workflow Designer are enabled the same way as other components, through capability activation, and cover two scenarios: workflow generation and workflow explanation.

1. Enable the Conversational AI capability for Workflow Designer.
2. Confirm that workflow administrators can access both the workflow proposal generator and the workflow explanation feature.

Workflow Generation

• A user describes the required workflow in natural language; the request is passed to the planner and generator services.
• The generator only produces components exposed through supported Workflow Designer tools. It does not invent unsupported triggers, actions, conditions, or property values, and ambiguous or unresolved references are not silently mapped to arbitrary values.
• The completed proposal transfers into Workflow Designer, remains fully editable, and is not automatically activated or executed.
• If generation fails, no partially configured active workflow is created.
• Results are returned after processing completes; streaming is not supported in this Technical Preview.

Workflow Explanation

• A user can request a plain-language explanation of an existing workflow from a supported Workflow Designer view.
• The explanation is based on read-only analysis. It does not execute, modify, save, publish, activate, deactivate, or create a new version of the workflow.
• The explanation covers what starts the workflow, which records or users it applies to, and what outcome it produces, using available display names and localized labels instead of technical identifiers where supported.
• Unsupported or unrecognized elements are clearly indicated where they materially affect interpretation, and the explanation does not invent behavior not present in the workflow definition.
• Access to the explanation respects the user's existing permission to view the workflow.

Known Limitations of This Preview

As a Technical Preview, Conversational AI does not yet include the following:

• Autonomous execution of AI-generated actions without user confirmation
• Proactive notifications or engagement through Microsoft Teams
• Live-agent handover from an AI conversation
• Remediation script execution or reporting capabilities in Microsoft Teams
• Complex service-request completion, including bundle or option selection, multi-recipient requests, field dependency handling, and full mandatory-field validation
• Conversational reporting or dashboards
• Personalized suggestions, recently viewed items, search history, or conversation history (including "last conversation" widgets) in the Self Service Portal
• Quick-action suggestions such as "Solve a problem," "Make a request," "Track my items," or "Show pending approvals"
• Voice interaction
• Follow-up refinement conversation, automatic activation or execution, and optimization, refactoring, risk, or runtime analysis in Workflow Designer
• Full support for all custom or third-party Workflow Designer components
• Streaming responses

Use Case: Requesting Support Through the Self Service Portal

The following scenario illustrates how an end user experiences Conversational AI when it is enabled for the Self Service Portal.

1. An employee opens the Self Service Portal to report a problem with their laptop. Instead of the previous Global Search field, the landing page now shows a prominent conversational entry point.
2. The employee types a natural-language description of the issue, for example, describing that their laptop is running unusually slowly. There is no need to know which service category, ticket type, or catalog item this falls under.
3. The request is sent through the portal's connection to Matrix42 Intelligence. The portal displays a response within the same conversational experience, without navigating the employee away from the page.
4. If the issue matches a supported simple scenario, the response provides direct guidance the employee can try immediately. If the issue needs to be tracked, the conversation results in a Matrix42 record being created in the employee's context, and the assistant confirms this with a reference the employee can use to follow up.