Agentic applications and orchestration

Filter GenAI consumption by model and key

The Consumption page for your GenAI Resource Packs now includes Model and Key filters. Select a specific model to see how much it is consuming or select a resource key to break usage down by the app or integration behind it. The filters apply across all consumption charts, including GenAI Units, input and output tokens, and API calls.

What this means for you: Pinpoint exactly what is driving consumption on a resource. If you connect several apps or models to the same GenAI Resource Pack, you can now attribute usage precisely, investigate unexpected spikes, and manage your GenAI Unit spend with far more confidence.

Event sub-processes: Now triggered by timers in Mendix Workflow

You’re familiar with event sub-processes in Mendix Workflow, useful for handling exceptions or optional paths during a workflow’s lifecycle. Until now, these could only be initiated by a message or notification.

With this release, we’re extending Event Sub-process capabilities to include Timer Start Events. You can now have time-based logic directly within event sub-processes. Set one to trigger after a specific duration has passed, or at a precise point in time—giving you fine-grained control over how and when your workflows respond.

Common use cases:

• Automated escalation: If the “main” normal flow isn’t completed within X-hours, trigger an event sub-process to escalate it.
• Scheduled follow-ups: Automatically initiate a follow-up action Y-days after the start of the workflow.
• SLA monitoring: Start an event sub-process if a service level agreement is about to be breached (normal flow should be completed within Z weeks), allowing for preemptive action

The benefit of this extended capability is that it simplifies the implementation of time-dependent processes. No need for external scheduled events or complex workarounds to introduce delays or time-based triggers within your exception handling or optional flows.

Agents Kit 2.1

Azure AI Foundry support in Agent Editor

The Agent Editor now supports Azure AI Foundry as a platform for text generation models, giving you the freedom to build AI agents using your existing Azure infrastructure. Select Azure AI as the provider from the Configuration panel and connect your agent to Azure’s models—bringing you more flexibility in how you power your AI agents.

AI-Assisted development for Agentic Applications

ASkill.mdfile has been added to the GenAI Commons, Agent Commons, Conversational UI, and Mendix Cloud GenAI Connector modules. This enables tailored Maia support, allowing the assistant to better understand and guide users within the context of these specific modules.

Find unhealthy agents fast

Previously, associated but missing objects would be identified inside of the agent version details only. In addition, a health check is now performed as a scheduled event on the agent overview page, providing developers a clear overview of whether any version used for production is missing a connected model, knowledge base or microflow tool. Also, the health check is used for validation when marking a version as In Use. The check is available in Agent Commons USE ME folder and can be reused wherever it suits the application best (e.g. inside of an After Start Up action).

Also, we upgraded all core modules to 11.12.1 and brought the GenAI Showcase application up to date with these latest features, so they can be tested in there.

Agentic development

Module updates while retaining customizations

Previously when a module was customized and an update was applied, all customizations would be discarded. This resulted in time-consuming updates and error-prone redoing of the work.

From Mendix 11.13, it is now possible to update your modules, while retaining your local customizations. Select Update from file from the App Explorer, and provide both the original .mpk as well as the .mpk with the update. The update triggers the merge algorithm to combine the changes you may have made locally, with whatever comes in from the server. And in case of a conflict, like shown below, you can choose which changes to retain.

This first step only works for modules created in Mendix 11.12 or above, but in the next release we’ll add support for modules created before that. We will also backport these improvements to 11.12 LTS which will significantly simplify the upgrade from Mendix 10!

Maia Make: Microflow layout improvements

We have made significant layout improvements when generating more advanced microflows with Maia. By optimizing microflow skill files to better leverage Sonnet 4.6+ thinking and simplifying microflow Model API schemas, we now produce cleaner and more complete microflows. This particularly benefits complex microflows, including those with loops and conditional sequence flows, resulting in a significantly improved layout and overall quality.

Maia Make: Expression errors known to Maia

Maia can now access full expression error details. This allows proactive resolution of validation errors to due exact issues being known. This means the user focus shifts to logic over debugging, the result being improved generation quality and faster development.

The last Dojo widgets go React

With React compatible alternatives now available for the Image Cropper and Pusher, all platform-supported widgets have been migrated or have a React-ready equivalent. This closes out the Dojo migration, so you can build and modernize apps on a fully React-based foundation.

Image Cropper

The Image Cropper is now available as a React compatible widget, replacing the older Image Crop module with full functionality intact.

End users can crop, zoom, rotate, and convert images to grayscale directly in the app, with rectangular or circular shapes and preset or custom aspect ratios. Changes save back to the image attribute automatically as the user works, so there is no separate logic for applying it and no need for external editing tools.

Pusher

The Pusher module now has a React compatible version, letting the server trigger a microflow or nanoflow in the client without any user action. This enables real time updates when data changes in another user session, so you can build features like live refreshes and notifications without falling back on constant page refreshes or polling.

Enterprise-grade platform

More control over who can maintain your Marketplace components

You can now decide who is able to maintain your company’s components in the Mendix Marketplace. Maintaining a component means editing it, releasing new versions, deprecating it, or unpublishing it. Until now, if a component was not part of a component group, any user with a Mendix account in your company could do all of this.

This change gives you clearer control over your components. A component can now be maintained only by the people who should be involved: the user who uploaded it, the developers chosen as collaborators during onboarding, the members of a component group that includes it, and your company’s Mendix admins. This helps prevent unwanted changes to components by people who have no connection to them.

For example, if your team shares a set of reusable modules, you can create a component group and add the right people to it. Everyone in that group can maintain those components together, while the rest of your company cannot. You can organize groups in the way that fits you best, such as by team, product, or owner.

If you have many components and no component group yet, this is a good moment to set one up. Component groups let you match maintenance access to how your teams actually work.

As part of keeping the Mendix Marketplace clean, trustworthy, and aligned with open-source standards, we have also retired public components whose latest version only supported Mendix 7, which is no longer supported. You can find the complete list in our Marketplace release notes.

Data and integration

“Send Email” microflow activity (public beta)

Sending emails with Mendix just gets a lot easier. We’re excited to introduce a new, lightweight way to send emails directly from your microflows, now available in beta. With this activity, you don’t need to configure the full-fledged email connector module when you just need to send a simple email. This new Send Email activity is a simple solution, allowing you to easily configure your connection details and message content right within the activity. With support for templates for both subject and content, and the ability to send test emails during design time, you can review and verify your configuration with ease. What’s more, this activity also comes with Maia support out-of-the-box. You can ask Maia to set up the configuration and message contents for you.

Important Note: Please remember to never share your authentication credentials and secrets with Maia

Data Importer: Fully Studio Pro native and works with Maia

The Data Importer is now fully native within Studio Pro, meaning you no longer need a separate module to leverage its capabilities. Furthermore, in 11.13, it comes with Maia support. Simply provide Maia with an example row from your CSV file, and watch as it sets up the Data Importer and maps your CSV table into the necessary entities for you. This new capability reduces manual effort and accelerates your development process, making data integration even more seamless. In the future, you can attach the CSV file directly to Maia.

Ready to transform your next idea into an app?

• Kickstart your project by structuring your vision with Maia Plan.
• Download Mendix Studio Pro and use Maia Make to turn your plan into an app.
• Take a closer look at all the features, improvements, fixes, and more in the Mendix 11.13 release notes.
• Don’t want to miss future updates? Subscribe to our blog to catch the latest news