Innatera introduces developer program for neuromorphic AI

Innatera has introduced a Developer Program that gives engineers, researchers, and hobbyists tools and resources to develop, test, and deploy neuromorphic AI applications.

The program extends a previously announced initiative. Hardware is available through DigiKey in the European Union, and availability is expected to reach additional countries by the end of 2026. Innatera was incorporated in 2018 as a spin-off from the Delft University of Technology and has raised EUR 55m from European deep-tech investors.

The program centers on Pulsar, a Spiking Neural Processor that delivers ultra-low-power intelligence at the point where sensor data is captured. Listed applications include human presence detection, keyword spotting, audio scene classification, and predictive maintenance across consumer and industrial devices. Developers work with the Innatera SDK, which contains Talamo for AI application development and compilation and Spine for embedded development and execution on Pulsar.

The Developer Portal functions as the central hub for software, documentation, application examples, community discussion, and support. Talamo supports a Python-based workflow with PyTorch integration, letting developers build sensor-processing pipelines, train neural networks, and simulate applications before deployment to hardware. Evaluation hardware consists of two boards: IgnisCore, the processing module, and SensForge, which connects to it to add sensors and connectivity. Developers can register through the portal.

“Neuromorphic computing will advance through what developers build with it,” said Sumeet Kumar, CEO of Innatera. “Our responsibility is to put this technology into their hands with tools they can use confidently. By bringing evaluation hardware together with familiar software workflows, technical guidance, and a developer community, we're helping more engineers move from experimentation to applications that solve real-world challenges.”

The company has set a goal of bringing intelligence to a billion sensors by 2030.