Intelligence for Early Adopters 26.1.x
Table of Contents
Matrix42 Intelligence is entering its next stage with a fundamentally improved AI experience for Matrix42 Enterprise.
Selected customers are already using the new experience in their production environments, where it replaces the previous AI approach. Initial customer feedback has been very positive and indicates a significant improvement compared with the previous experience.
Early Adopter customers receive dedicated 26.1.x Early Adopter builds, allowing them to benefit from prioritized fixes and improvements more quickly. Updated builds are planned on a weekly basis where possible and incorporate both findings from our own validation and feedback from customers using Matrix42 Intelligence in real production scenarios.
Early Adopter Does not Mean Experimental Quality
The Early Adopter program should not be understood as providing an experimental or low-quality version of Matrix42 Intelligence.
The new Intelligence experience is already replacing the previous AI approach for a selected group of customers in production. Feedback from these customers shows that the new capabilities provide a substantially better experience and greater value in everyday service-management scenarios.
The purpose of the Early Adopter program is therefore not to determine whether the new approach works. It enables Matrix42 to work closely with selected customers during the rollout, identify real-world improvement opportunities faster, and deliver prioritized improvements without waiting for the next regular product release.
As usage grows, we expect to identify additional opportunities to further improve quality, usability, relevance, and performance. Dedicated 26.1.x Early Adopter builds enable us to respond to these findings quickly.
Dedicated 26.1.x Early Adopter Builds
Early Adopter customers receive dedicated Matrix42 Enterprise 26.1.x builds containing updates related to Matrix42 Intelligence.
We plan to provide updated builds on a weekly basis where possible.
These builds may contain:
- Fixes for identified issues
- Improvements based on Early Adopter feedback
- Quality and usability improvements identified through our own validation
- Improvements to AI response relevance and consistency
- Improvements to Native AI Functions
- Improvements to the Agentic Conversation experience
- Performance and stability improvements
Feedback from Early Adopters is continuously reviewed together with issues and improvement opportunities identified by Matrix42 and used to prioritize upcoming builds.
The objective is to improve Matrix42 Intelligence incrementally and rapidly rather than requiring customers to wait for the next regular Matrix42 Enterprise release.
A New AI Foundation for Matrix42 Intelligence
The change is not limited to a new user interface.
As part of this evolution, Matrix42 has also replaced the underlying AI approach used to execute Native AI Functions.
The new AI engine and orchestration approach are designed to provide:
- More relevant and useful AI results
- Better use of the available business and ticket context
- More consistent behavior across AI capabilities
- A stronger foundation for agentic capabilities
- Better integration between conversational AI, Native AI Functions, and Matrix42 Enterprise data and processes
This means that existing use cases such as classification, summarization, finding related information, and finding solutions benefit not only from improved integration into the product but also from the new underlying Intelligence engine.
The new Agentic Conversation experience and Native AI Functions therefore share the same broader evolution of the Matrix42 Intelligence platform rather than representing only a visual replacement of the previous AI functionality.
Agentic Conversation Experience
The previous AI Chat has been replaced with a new conversational experience built on fundamentally different technology.
Instead of operating like as a standalone AI tool, the new experience is designed to understand the context in which an agent is currently working.
It is designed to:
- Understand what the agent is working on without requiring the context to be repeatedly explained
- Provide responses that can be acted on immediately, rather than only information that needs to be interpreted
- Work directly with data and processes available in Matrix42 Enterprise
- Take supported actions on behalf of the agent where enabled
- Make interaction with AI feel like a natural conversation rather than another tool that needs to be operated
Instead of simply moving the previous AI Chat into another panel, Matrix42 has replaced it with an Agentic Conversation Experience centered around the agent's actual work context.
Native AI Functions
The previous AI Assist Bar has also been replaced by Native AI Functions.
Rather than grouping AI capabilities in a separate toolbar, Native AI Functions are integrated directly into Matrix42 Enterprise as standard actions and contextual elements.
AI capabilities can therefore appear exactly where agents are already working, for example as actions, suggestions, badges, or capabilities within a ticket preview.
The following Native AI Functions are available:
- Classification: suggests category, impact, and urgency directly on the ticket
- Summarize Ticket
- Knowledge Article Drafting
- Find Similar Tickets and Knowledge Articles
- Find Solution
- Text Enhancement
- Translation and Localization
Together with their deeper integration into the user workflow, these functions now use the new Matrix42 Intelligence AI approach, designed to provide better and more context-aware results than the previous implementation.
How this Looks in Everyday Work
The objective is to make AI part of the agent's normal workflow instead of requiring agents to switch to a separate AI application.
For example:
- Agents can see a ticket summary directly within the ticket context.
- Classification suggestions can appear directly on a ticket.
- Agents can find similar tickets and Knowledge Base articles without leaving their current workflow.
- Find Solution can generate a proposed solution using the available ticket and knowledge context.
- A generated solution can be used directly when sending an email or closing a ticket, reducing manual copy and paste.
- Agents can create a Knowledge Article draft based on existing ticket information.
- Agents can ask Matrix42 Intelligence for assistance in plain language through the new conversational experience.
- The conversational experience can understand the context of the agent's current work instead of requiring that information to be entered again.
More Flexibility for Administrators
Native AI Functions use the standard Matrix42 Enterprise configuration experience.
Administrators can:
- Manage access to available AI Functions
- Configure where functions are shown
- Define the appropriate target user groups
- Customize existing AI Functions
- Introduce customer-specific AI Functions
This provides administrators with a familiar way to adapt AI capabilities to the way their organization works without introducing a separate AI-specific administration model.
Continuous Improvement during the Early Adopter Phase
The Early Adopter rollout is intentionally iterative.
Based on our own validation and initial customer feedback, we have already identified a number of issues and opportunities for further improvement. These are continuously reviewed and prioritized.
Where possible, prioritized improvements will be delivered through the weekly 26.1.x Early Adopter builds.
Current areas of continuous improvement include:
- AI Search relevance and ranking
- Find Similar Tickets and Knowledge Articles
- Find Solution quality and consistency
- Context used by AI Functions
- Conversational context handling
- Native AI Function usability
- Performance
- Stability
- User experience
- Customer-specific configuration scenarios
This does not mean that these capabilities are considered unfinished or unsuitable for production. It reflects our approach of using real customer feedback to continue improving an already significantly enhanced experience.
Feedback is Part of the Early Adopter Approach
Feedback from Early Adopters is an important input into this process.
We encourage customers to collect feedback from both administrators and agents as they work with the new Matrix42 Intelligence capabilities.
Feedback is evaluated together with already identified product issues and improvement opportunities and is used to prioritize upcoming Early Adopter builds.
We have already identified a set of issues and improvements that we are actively prioritizing. The current list and status can be followed here:
Solved problems and improvements
List of solved problems and minor improvements for Matrix42 Enterprise 26.1.x, including all updates.
Important: Update Matrix42 Intelligence Extensions
To ensure compatibility with the latest Matrix42 Intelligence capabilities, customers must update the relevant Matrix42 Intelligence extensions in addition to installing the latest Matrix42 ESM Platform release.
Please make sure that the latest available versions of the following extensions are installed:
- AI-Native Workflow Designer
- AI Assist for Service Desk
- AI Intelligence Foundation
- Agentic Enterprise Assistant
Updating only the ESM Platform is not sufficient. The latest extension versions contain required fixes, improvements, and compatibility updates for the current Matrix42 Intelligence experience.
We therefore recommend verifying and updating all four extensions after upgrading the ESM Platform.
Release 26.1.0.3740 (01.09.2026)
| ID | Type | Area | Description |
|---|---|---|---|
| 596768 | Improvement | AI Functions / Find Solution | Improved the consistency of Find Solution results. The tools an AI Function may use are now configured explicitly on the AI Function implementation, so every agent runs the function with the same tool set. This addresses proposed responses that differed between agents or referred to content that was not on the ticket, such as attachments the customer never provided. |
| 599284 | Improvement | Ticket Journal | Refined the ticket journal introduced in build 3723 based on early-adopter feedback. Entry origin and category now have distinct colors and icons, the entry content has tighter spacing, and the Translate and Enhance actions carry tooltips and accessible labels. |
| 600432 | Issue | Assistant / Find Solution | Fixed an issue where the Assistant got stuck when Find Solution did not find a Knowledge Base article to return. The conversation now completes normally and the agent can continue, for example by searching for an article manually. |
| 598854 | Issue | Personalization / AI Output Language | Fixed the AI output language showing Current language in Personalization while AI Functions actually answered in English. When no output language is personalized, the setting now shows None and AI Functions, including Summarize, use the language from the user's profile. |
| 597424 | Issue | AI Functions – Translation | Fixed an error when running Translate on a ticket description or a Knowledge Base article. Instead of the translated text, users previously received a PromptExecutionFailed exception. |
| 598023 | Issue | Service Catalog / E-mail Notifications | Fixed an issue where every recipient's e-mail address was visible to all recipients when a notification about a service state change was sent. The Send Mail compliance rule now supports a BCC recipient option, which the notification uses. |
Release 26.1.0.3723 (25.08.2026)
| ID | Type | Area | Description |
|---|---|---|---|
| 597241 | Improvement | AI Functions | Improved the Apply experience for AI Function results. After a value has been applied, the wizard remains visible but the Apply action is disabled and clearly indicates that the result has already been executed, making it easier to understand which value was applied. |
| 594715 | Improvement | Knowledge Article Drafting | Knowledge Article Drafting now uses the selected system template when generating a draft, helping ensure that AI-generated articles follow the configured Knowledge Base structure. |
| 593651 | Improvement | AI Functions / Ticket Preview / Journal | Added a Summarize Text AI Function for long ticket descriptions and journal entries. Agents can quickly generate a concise summary of lengthy content directly from the ticket or journal entry, with the result respecting their AI personalization settings. |
| 594355 | Improvement | AI Functions | Improved protection against unnecessary repeated AI Function executions. While an AI Function is running, duplicate executions are prevented. If the same function is run again without relevant context changes, users are warned that another execution will consume additional Action Credits and may generate a different result. |
| N/a | Improvement | Ticket Journal | Improved the Ticket Journal for Agents with entry categorization, filtering, and keyword search. Users can distinguish between customer-facing updates, internal notes, changes, escalations, system/automation entries, and other journal types, filter between AI-generated and human-created entries, and search journal content by keyword. AI-generated entries are also visually identified, making long ticket histories easier to scan, audit, and navigate. |
| 598208 | Improvement | Self Service / Ticket Journal | Introduced a new Simplified Journal view for end users in Ticket, Incident, and Service Request previews. The simplified mode removes technical controls and filtering, uses clearer entry presentation with avatars/icons and timestamps, and provides a cleaner, easier-to-read history focused on information relevant to end users. |
| 594545 | Issue | AI Search | Improved AI Search result ranking so that the most relevant results are more consistently shown at the top of the result list. |
Release 26.1.0.3687 (18.08.2026)
| ID | Type | Area | Description |
|---|---|---|---|
| 596335 | Issue | Assistant | Fixed an issue where the ticket or AI conversation could close unexpectedly even when Assistant was configured to remain open. |
| 596505 | Issue | Native AI Functions | Removed an unnecessary Close button from AI Functions to provide a clearer and more consistent user experience. |
| 596196 | Improvement | Assistant / AI Actions | AI Actions can now be accessed directly from the conversation while working on a ticket, without leaving the current ticket context. |
| 597528 | Improvement | AI Search | Updated the AI Search hint in the application header from “Ask anything” to “Search…”, making it clearer when to use Search versus Assistant. |
| 594704 | Issue | Assistant | Fixed inconsistent font styling in Assistant so that the integrated AI conversation now follows the typography and visual styling of Matrix42 Enterprise. |
| 596339 | Issue | Assistant | Fixed an issue where new conversations continued to be shown as “New Chat”. Conversation names are now updated automatically after the AI responds, making chats easier to identify and navigate. |
| 595217 | Improvement | Assistant | Added a Full Screen mode for Assistant, giving users more space for longer or more complex AI conversations. Full Screen can be opened from other display modes and can also be saved as the preferred default experience. |
| 596467 | Issue | Assistant | Fixed an issue where tickets created through Assistant could receive an INC number instead of the expected TCK number. Ticket numbering rules are now preserved correctly, including when the created record represents an Incident. |
| 594547 | Issue | AI Search | Improved AI Search performance in scenarios where searches could take longer than expected, reducing response times when retrieving relevant information. |
| 596474 | Issue | M42 Assist | Fixed an issue where conversations could fail when AI responses contained very large content, such as images or extensive tool results. M42 Assist now handles these larger responses more reliably instead of interrupting the conversation. |
Known Issues and Improvements
| ID | Area | Known issue | Customer impact |
|---|---|---|---|
| 598103 | AI Functions / AI Agents | AI Functions or AI Agents may fail when executed under a user account that does not have an email address configured. | AI-powered actions, including automated ticket processing, may return an error instead of completing when the executing user has no email address defined. |
| 594530 | AI Search | Search results may include irrelevant Knowledge Base articles. | Some returned articles may not sufficiently match the user’s query or ticket context. |
| 594550 | AI Search / Assistant | Not all Knowledge Base articles accessible to the user may be found or considered by AI. | Search or conversational AI may occasionally omit articles that the user otherwise has permission to access. |
| 596614 | AI Functions | The Find Related AI action may fail in some scenarios. | Instead of related tickets or articles, the action can return an error. |
| 596474 | Assistant | Long conversation messages can cause the conversation to fail. | Very large content, such as large images or extensive tool output, may exceed technical message-size limits and interrupt the conversation. |
| 594551 | Conversation History | Conversations started from a ticket do not include the Ticket ID in the conversation name. | Conversation history can be harder to associate with the originating ticket. |
| 596334 | Assistant | A renamed conversation may revert to its previous name. | Custom conversation names are not always persisted. |
| 594243 | AI Search | Repeating the same search may return different results. | Users may see some variation in search results when running the same query multiple times. |
| 594548 | AI Search / Find Related | Custom Knowledge Base fields are not fully considered when finding relevant articles. | Information stored only in customized KB fields may not influence search or related-article results as expected. |
| 596185 | Assistant | In some cases, the Assistant may report an action as completed even though it was not successfully executed. | Users should verify the resulting record or change when an AI-executed action reports successful completion. |
| 596186 | Assistant | A failed ticket action may leave partial changes and can create duplicate journal entries when retried. | Retrying an unsuccessful action can occasionally result in duplicated journal entries or partially applied changes. |
| 596464 | Assistant | Email notifications may not be triggered when the Assistant creates or closes a ticket. | Expected ticket notifications may not be sent for tickets created or closed through the Assistant. |
| 596465 | Assistant | Some ticket update actions may fail for non-administrator users. | Users without administrative permissions may receive an error when the Assistant attempts to update certain ticket fields. |
| 596784 | Assistant | The Assistant may fail to open for certain Service Request items. | Users may receive an error instead of the Assistant when working with affected Service Requests. |
| 597123 | Assistant / Self Service | Links to services suggested by the Assistant may open the Self Service Portal home page instead of the specific service or form. | Users may need to search for the suggested service manually after following the link. |
| 596189 | Assistant | Long-running AI actions may provide limited progress information and can eventually fail. | Users may see an extended waiting state without clear progress before the action completes or returns an error. |
| 596463 | Assistant | Mandatory fields may not always be enforced when a ticket is created through the Assistant. | Tickets created by the Assistant can in some scenarios be created without information that is normally mandatory. |
| 596187 | Assistant | Ticket changes performed by the Assistant may not immediately appear in the currently open ticket view. | Users may need to reopen or refresh the ticket to see changes made by the Assistant. |
| 596190 | Assistant | Forwarding a ticket may fail when the target user is identified by their full name. | Some ticket forwarding requests performed through the Assistant may return an error. |
| 597175 | Knowledge Article Drafting | Some Knowledge Article templates may not be available for selection because of template-type filtering. | Users may be unable to use an otherwise applicable template when drafting a Knowledge Article. |
| 597195 | Knowledge Article Drafting | AI-generated Knowledge Article drafts may not fully preserve formatting from the selected template. | Elements such as icons or tables from the template may be missing or changed in the generated draft. |
| 597122 | Assistant | An ongoing AI response cannot currently be interrupted by the user. | Users must wait until the current response finishes before starting another interaction. |
| 596469 | Assistant | When the Assistant creates a ticket, a relevant asset identified from the conversation may not be automatically linked to the ticket. | Users may need to manually add the affected or relevant asset after the ticket is created. |
| 596942 | Assistant | Progress information may be missing for some steps performed by the Assistant. | During multi-step operations, users may not always see information about what the Assistant is currently doing. |
| 594715 | Knowledge Article Drafting | Knowledge Article drafts do not always use the selected system template as expected. | Generated articles may require additional manual formatting or adjustments to match the intended Knowledge Article template. |
| 594355 | AI Functions | An AI Function may be executed again unnecessarily when users revisit or interact with its result. | Users may experience additional waiting time and, depending on configuration, unnecessary AI executions. |
| 597241 | AI Functions | Some controls and preview behavior during AI Function execution are not yet fully consistent with the intended workflow. | Users may see an unnecessary Cancel option, an Apply action that does not clearly indicate that it has already been executed, or reduced preview space when the Assistant panel is configured to open automatically. |
Recommended next steps
We recommend that your Matrix42 administrator:
- Reviews the Native AI Functions now available
- Verifies access permissions and target user groups
- Informs agents about the new Agentic Conversation experience
- Informs agents that the previous AI Assist Bar has been replaced by Native AI Functions
- Reviews customized ticket previews and layouts where applicable
- Updates the environment to the latest provided 26.1.x Early Adopter build
- Reviews the Known Issues and Improvements list when a new build is provided
- Collects feedback from agents using the new experience
- Shares relevant feedback with Matrix42
Documentation
For information about using and configuring the new capabilities, see:
Early Adopter Build Cadence
Matrix42 plans to provide updated 26.1.x Early Adopter builds on a weekly basis where possible. Each build can include prioritized fixes, quality improvements, and enhancements resulting from both Matrix42 validation and customer feedback.
This approach enables Early Adopter customers to benefit from improvements significantly faster while helping Matrix42 continuously refine the new Intelligence experience based on real production usage.