Train and deploy your own AI models in Arduino® App Lab – now fully integrated with Edge Impulse!

We’re excited to introduce a powerful new capability in Arduino App Lab: native integration with Edge Impulse for training and deploying your own custom AI models.

This update makes it easier than ever to bring machine learning into your App Lab projects. By connecting directly to Edge Impulse Studio, you can now train models on your own data and seamlessly use them inside Arduino App Lab, confidently building applications that tackle specialized tasks, unique datasets, and project-specific goals.

While Arduino App Lab has always made it simple to get started with pre-built AI examples, we knew there was a strong desire for a smoother way to develop and use custom AI models. With this release, that experience is now fully integrated and guided through the UI, making custom AI a natural part of the Arduino App Lab workflow.

A smooth path from data to deployment

Arduino App Lab now works hand-in-hand with Edge Impulse, removing friction and letting you focus on building great projects instead of wiring tools together. At a high level, the flow looks like this:

• Open an AI-powered example in Arduino App Lab or create your own
• Navigate to Bricks > AI Models and select Train new AI model

_{^{The new “Train AI model” link to Edge Impulse Studio from Object Detection Brick}}

• Sign in with your Arduino account and connect to Edge Impulse

• Create and train your model in Edge Impulse Studio

^{Example of an Edge Impulse custom ML model for detecting apples and bananas with the Arduino® UNO™ Q}

• Return to Arduino App Lab to find your trained model ready to use
• Configure your Bricks, install the model on your board, and deploy

You can also work with multiple impulses, switch between models, and manage everything directly from the Arduino App Lab UI. For a detailed walkthrough, including best practices and tutorials, refer to the full documentation available at this link. 

See it in action

To show how this new integration works in practice, we’ve created a video that walks through the full flow using a real custom model example.

As a developer, Louis Moreau from Edge Impulse, is no stranger to rubber ducks. Like many of us, he keeps a small army of them around to help think through problems and debug code. Naturally, when the time came to try out custom AI models in Arduino App Lab, he decided to have a bit of fun with it!

In the video, Moreau builds a custom vision model to detect rubber ducks, using Edge Impulse directly from Arduino App Lab.

Get started with edge AI in Arduino App Lab

This integration brings the flexibility of Edge Impulse together with the simplicity of Arduino App Lab. Now you can experiment, prototype, or build something truly unique leveraging custom machine learning – now easier to define, manage, and deploy without breaking your creative flow.

Ready to begin your edge AI journey? Whether you’re experimenting for the first time or building a more advanced project, these resources will help you get up and running quickly: