Simantic, a Y Combinator (F26) startup founded by ex-firmware engineers from Neuralink and Tesla, is building a simulation layer that enables AI agents to test and validate firmware without physical hardware. The founders, Ahnaf and Seungmin, developed the platform to address the slow, manual validation processes traditionally required in firmware development. With Simantic, engineers and AI agents can run virtual devboards via a single API call, perform AI-powered code reviews backed by simulation-based verification, and integrate CI/CD pipelines that validate firmware against simulated microcontrollers on every commit.
The tool is already being used in high-stakes domains including flight controllers, electric vehicles, robotics, and satellites. One early-stage space startup used Simantic to test satellite firmware on non-existent hardware, saving over $100,000 in development costs and reducing testing time significantly. Unlike open-source emulation frameworks like Renode and QEMU, which Simantic initially relied on, the team built a custom simulation engine to overcome limitations in speed, accuracy, and flexibility. They found Renode too slow, constrained by its Dotnet foundation, and overly bloated with unused features. Simantic’s architecture is optimized for real-time feedback and precision, especially critical in safety-sensitive applications where firmware errors can cost lives or millions in hardware.
Source: Hacker News · Summarized by HeadlinesBriefing