Blog

No Magic, Just Mendix: Making AI Accessible with the Mendix ML Kit 

In the ever-evolving landscape of technology, artificial intelligence (AI) and machine learning (ML) continue to revolutionize how we build and use software. But for many developers, integrating ML into applications remains complex and time-consuming. 
That’s where the Mendix Machine Learning Kit comes in — a game-changing toolkit that simplifies the integration of machine learning into Mendix applications, enabling developers to enhance apps with intelligent automation, real-time insights, and smarter decision-making.

Why Machine Learning? 

At its core, machine learning is a subset of AI that empowers systems to learn from data and make predictions or decisions without being explicitly programmed. ML models sift through massive data sets, identify patterns, and evolve over time — all autonomously. 
But traditionally, embedding these models into applications has required significant technical effort. That’s what the Mendix ML Kit seeks to change. 

Introducing the Mendix Machine Learning Kit 

The Mendix ML Kit allows developers to integrate pre-trained machine learning models directly into their Mendix applications using the Open Neural Network Exchange (ONNX) format. Whether you’re looking to automate workflows, personalize services, or generate real-time predictions, the ML Kit makes it easy. 

Some standout features include: 

• Seamless ONNX Runtime integration 
• Pre- and post-processing capabilities 
• Compatibility with any ML framework (e.g., Scikit-learn, TensorFlow) 
• Support for ensemble learning, batch inference, and cascaded inference 

Under the Hood: Pre- & Post-Processing 

Before a machine learning model can deliver actionable insights, it needs to understand the data it’s working with—and present its results in a meaningful way. That’s where pre- and post-processing come in. The Mendix ML Kit covers these steps with dedicated Java actions in microflows, making it easy to integrate smart data handling into your apps.

Pre-Processing prepares raw input data for the model. This includes:

• Data transformation: Converts diverse inputs—like text, images, or categories—into numerical values the model can interpret
• Normalization and scaling: Ensures consistent data ranges, preventing skewed results
• Feature engineering: Identifies and enhances the most relevant characteristics in your data to improve model accuracy
• Encoding: Translates categorical data into machine-readable formats without losing meaning

Post-Processing turns raw model predictions into real-world value:

• Decoding predictions: Converts complex output into understandable terms for end users
• Classification and thresholding: Maps probabilistic outcomes to clear, actionable categories
• Business logic integration: Tailors results to match your organization’s unique requirements and decision rules
• User-friendly formatting: Displays output in intuitive formats such as visual dashboards, structured tables, or plain-language summaries

By automating these steps, the ML Kit bridges the gap between advanced machine learning and everyday business applications

Real-World Use Case: News Headline Categorization 

Let’s say you want to categorize news headlines — is it politics, finance, or tech? Using Scikit-learn, you can train a Random Forest model and export it in ONNX format. Then: 

1. Pre-process with TF-IDF vectorization in java actions in microflows
2. Embed the model into Mendix Studio Pro
3. Run the data and get predictions (e.g., “Tesla car crash…” → category: “automobile”) 
4. Post-process the output to translate model predictions into categoriesin java in microflow actions

All of this happens within the Mendix environment — no need for external hosting or third-party deployment pipelines. 

Why Use the Mendix ML Kit?

Bringing machine learning into your applications doesn’t have to mean long development cycles, complex integrations, or relying on external services. The Mendix ML Kit streamlines the entire process, offering a fast, intuitive way to embed intelligent features into your low-code apps.
By enabling models to run directly within Mendix, the ML Kit dramatically reduces time-to-market. What once took weeks of backend setup and testing can now be accomplished in days—or even hours—thanks to pre-built microflow actions and simple deployment steps.
Integration is seamless, requiring no advanced configuration or custom middleware. Whether you’re connecting a pre-trained model or integrating one from an external service, the ML Kit handles the heavy lifting behind the scenes, so you can focus on building value into your app.
And because models run inside the Mendix environment, performance is optimized. There’s no need to call external APIs, which means lower latency, better reliability, and a smoother experience for your users.
In short, the Mendix ML Kit makes AI development accessible. You don’t need to be a machine learning expert to build apps that recognize images, classify content, or personalize user interactions. Whether you’re automating inspections or building smarter forms, Mendix makes it as simple as drag-and-drop.

No magic. Just Mendix.