- Embedd, a London-based startup founded by Ukrainian entrepreneurs, has raised $2.7 million in pre-seed funding led by Seedcamp.
- Its AI platform cuts the software integration work behind robots, cars, drones and medical devices from months down to weeks.
- The startup is already working with Microchip Technology, one of the world’s largest semiconductor makers.
Embedd, a London-based startup building software infrastructure for robots, cars, drones and medical devices, has raised $2.7 million in pre-seed funding led by Seedcamp.
Cocoa, Connect Ventures, 2100 Ventures, Vesna Capital, U.ventures, Underline Ventures, Common Magic, and Roosh Ventures also joined the round.
Every intelligent machine runs on dozens of chips that don’t share a common language, and engineers currently spend four to six months hand-writing the code that lets them talk to each other. Embedd’s platform cuts that down to around three weeks, an eightfold reduction, and says it can deliver production-ready software for a given chip up to six times faster than the manual process.
“The next wave of AI will power factories, vehicles, robots and critical infrastructure, but today, every change in hardware creates huge complexity for software teams, and that friction is already massively slowing innovation,” said Michael Lazarenko, co-founder and CEO of Embedd.
A software gap hiding behind the physical AI boom
The money is arriving at a busy moment for the category. Nearly $19 billion has gone into robotics and physical AI startups so far this year, with $16.3 billion of that landing in the first quarter alone. Most of it is chasing robots and foundation models. Embedd is chasing something less visible: the software layer that none of that hardware works without.
Lazarenko founded Embedd with Maxim Gorinov and Valentin Gololobov, Ukrainian entrepreneurs who ran a hardware company together before this one. Covid-era chip shortages forced them to keep re-sourcing components and rewriting firmware from scratch, and Russia’s invasion of Ukraine later took their production facility outright.
Watching that rewrite cycle repeat itself is what convinced them the integration problem would only get worse as AI moved off screens and into physical machines. The startup works by turning chip documentation into a digital twin of the hardware, rather than having an engineer read a datasheet line by line.
From hardware disruption to a new software company
It’s not the only one going after this. Bengaluru-based H2LooP raised $2 million from Speciale Invest and 3one4 Capital earlier this year for a similar hardware-aware AI layer that pairs specialised models with a knowledge graph of hardware specs and safety standards. H2LooP relies on AI coding agents that work across existing codebases; Embedd builds outward from a structured hardware model instead. Which approach chipmakers standardise on is still an open question.
Since launching commercially in April 2026, Embedd has signed on several semiconductor customers, most notably Microchip Technology, where it’s enabling support for the Zephyr real-time operating system.
“The competitive question in embedded is no longer whose silicon is fastest, it’s whose silicon is easiest to build on. Our work with Embedd is about meeting developers inside the software ecosystems they’ve already committed to, rather than asking them to come to ours,” said Rodger Richey, Microchip’s vice president of development systems and academic programs.
Embedd launched commercially in April 2026 and has since signed contracts with multiple semiconductor companies.
Carlos Eduardo Espinal, general partner at Seedcamp, points to the scale of the problem as the draw: “Embedd is tackling one of the fundamental challenges facing industries from robotics and manufacturing to healthcare and automotive, and we’re excited to back Michael and the team as they build the infrastructure underpinning the next generation of intelligent machines.”
Embedd plans to use the new funding to expand its platform and deepen partnerships with semiconductor companies. Its broader ambition is to make hardware easier to adopt across software ecosystems, reducing the time needed to bring new chips into products as physical AI becomes more widespread.
