SiMa.ai has launched Palette Neat, an open-source agentic development environment designed for Physical AI applications. The platform combines a Physical AI execution library with an agent workflow layer to help developers reduce complex application development timelines from months to days.

According to the company, Palette Neat allows developers to use natural-language commands to build and deploy Physical AI systems. The environment is designed to abstract low-level compute complexity, helping engineering teams focus on system-level differentiation across new and legacy applications.

Krishna Rangasayee, Founder and CEO, SiMa.ai, said the company is delivering what it describes as the industry’s first agentic development environment for Physical AI. He added that Palette Neat, along with the company’s pin-compatible system-on-module, enables developers to design systems in plain English and develop them in days, and in some cases, hours.

The company said Palette Neat can autonomously build and map applications directly to silicon, while allowing developers to reuse existing application code and preserve about 90% of legacy software investment. The platform is aimed at reducing the cost, time and engineering risk involved in switching hardware platforms.

When paired with SiMa.ai’s full-production Modalix MLSoC System-on-Module or its new PCIe companion card form factor, the platform is built to support Physical AI workloads across robotics, automotive, drones, industrial automation, aerospace and defence, smart vision and healthcare.

The Modalix SoM is designed to run multiple large language models alongside vision and sensor models under 10W. SiMa.ai said the module is built as a pin-compatible replacement for the incumbent NVIDIA SoM form factor, requiring no carrier board redesign.

The company said Palette Neat is available as an open-source platform on GitHub, with documentation available through its Developer Center. SiMa.ai has also announced a webinar on scaling Physical AI, scheduled for 30 June.