Edge AI Silicon Labs Integrates AI Assistants with Hardware and MLOps

Related Vendors

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

SIMTOS (Seoul International Machine Tool Show)

Silicon Labs is expanding its developer platform to include AI tools, an open-source initiative for Bluetooth LE, and integration with Databricks. At the industry event Pretzl Connect, the company demonstrated in detail how AI assistants will verify hardware specifications and bring edge AI models all the way down to the microcontroller.

The Simplicity AI SDK is now in public beta. It provides AI assistants with structured access to Silicon Labs’ SDKs, development tools, documentation, and connected hardware. The initial focus is on Bluetooth Low Energy. The latest press release lists GitHub Copilot, Cursor, and Codex as supported AI assistants.

Silicon Labs launched the public beta on September 22. The SDK does not include its own language model. Instead, developers use their existing AI assistants, while Silicon Labs provides the technical context for its hardware and software through MCP servers, tools, and development expertise.

Hardware Intent Checks Specifications for Conflicts

Another new feature is Simplicity Design Intelligence. The first tool, called Hardware Intent, is designed to consolidate various specifications for a project. These include product requirements, schematics, and other hardware documents. In its briefing, Silicon Labs also cited meeting notes and other formats as potential sources.

The system is designed to assist developers in configuring pins, peripherals, and software, and then verify that the implementation aligns with the original requirements. According to Silicon Labs, it can identify inconsistencies regarding pins, peripherals, or other hardware conditions. The goal is to identify such conflicts before a circuit board goes into production, if possible. An alpha version of Hardware Intent is scheduled for release in January 2027.

At the same time, Silicon Labs is opening up parts of its software development process more broadly to contributions from the community. Bluetooth Low Energy is the first area to be included. Developers can review sample applications on GitHub, report issues, and suggest changes via pull requests. Contributions that Silicon Labs accepts undergo internal review and testing processes and may subsequently be incorporated into future versions of the Simplicity SDK.

Databricks Workflow Extends All the Way to the Microcontroller

The third focus area is Edge AI. In collaboration with Databricks, Silicon Labs integrates the development and management of machine learning models with embedded target hardware. Models can be trained in the Databricks environment and then tested to determine whether they fit within the memory and computational limits of a specific Silicon Labs chip.

The ML Profiler plays a key role in this. Tamas Daranyi, Product Marketing Manager for AI/ML at Silicon Labs, explained during the briefing: “The Model Profiler understands the hardware; it knows how the model will run on the dedicated accelerator and provides you with a very detailed analysis of how fast it will be and how much memory it will require.”

The profiler can also analyze models in simulation mode without the target hardware connected. The embedded workflow described by Silicon Labs uses quantized TensorFlow Lite models. Among other things, the tools analyze memory requirements and computational overhead, helping developers identify where they need to adapt a model for the target hardware.

Through the Databricks integration, the approach goes beyond the one-time deployment of a model. Devices can send back operational and training data, which can then be reused for training. In the briefing, Silicon Labs also outlined the prospect of later distributing updated models to larger device fleets via IoT infrastructures. This creates a cycle of field data, retraining, hardware analysis, and updated edge models.