This London Startup Just Raised Millions to Unlock Physical AI

Jan de Vries · 2026-08-24

Let's talk about a problem that's quietly grinding innovation to a halt. Picture this: a brilliant new chip gets designed, one that could make a factory robot smarter or an autonomous vehicle safer. But before it can do anything useful, software engineers need to spend months—maybe years—deciphering thousands of pages of technical docs just to get the software to talk to the hardware. It's a massive bottleneck, and it's only getting worse as AI moves out of the cloud and into the physical things around us. That's exactly the wall Embedd is trying to tear down. This London-based startup just secured $2.7 million in pre-Seed funding to build what they call the software infrastructure for physical AI. The round was led by Seedcamp, with a whole squad of other investors joining in, including Cocoa, Connect Ventures, and 2100 Ventures. ### The Founders' Pain Point Became The Solution The story here is personal. The founders—Michael Lazarenko, Maxim Gorinov, and Valentin Gololobov—are Ukrainian tech entrepreneurs who learned this lesson the hard way. Their previous hardware company got hammered by the global chip shortage during COVID. Then, Russia's invasion of Ukraine completely disrupted their operations. Every time they had to find a new component or a different chip supplier, their team faced the same nightmare: rewriting all the software from scratch. They realized they weren't alone. This was a universal choke point. "Every change in hardware creates huge complexity for software teams," explains Lazarenko, Embedd's CEO. "That friction is already massively slowing innovation." They built Embedd to be the solution they wished they'd had. ### How Embedd's Digital Twins Cut Through The Chaos So, what does Embedd actually do? Think of it as an AI-powered translator for the world of chips. Their platform uses digital twins and AI agents to automate the grueling process of integrating a new piece of hardware into a software ecosystem. Here's the old way versus the new way: - **The Old Bottleneck:** Engineers manually read thousands of pages of documentation for each unique chip. They then hand-write all the low-level code needed for the software above to command the hardware below. It's slow, expensive, and prone to errors. - **Embedd's Automation:** The company creates a precise digital twin of the physical hardware. Their AI agents use this model to understand the chip's context and automatically generate the necessary integration code. It's like giving every chip a common language. The result? Embedd reports that its customers can now deliver production-ready software for new chips up to six times faster. That's not just a minor speed boost—it's a complete change in the development timeline. ### Why This Matters For The Future This isn't just about making life easier for programmers. Lazarenko points to the "next wave of AI" that will power everything from factories and vehicles to robots and critical infrastructure. If we're going to fill our world with intelligent machines, we can't have software teams stuck in documentation purgatory for every single component. Since its commercial launch, Embedd has already started signing contracts with semiconductor companies. One notable partner is Microchip Technology, where Embedd is helping to enable support for the Zephyr real-time operating system. "The promise of physical AI is enormous," Lazarenko adds, "but today's hardware fragmentation is slowing innovation." This new funding is fuel for Embedd to expand its platform and help more companies bridge that gap. It's a bet that smoothing out this fundamental friction will accelerate how quickly smart ideas become real, working machines in our world.