London's Embedd raises €2.3M pre-Seed to automate chip-to-software integration for physical AI — Seedcamp leads

Embedd (London) has raised €2.3 million ($2.7 million) in a pre-Seed round led by Seedcamp, with Cocoa, Connect Ventures, 2100 Ventures, Vesna Capital, U.ventures, Underline Ventures, Common Magic, and Roosh Ventures participating. The company commercially launched in April 2026 and has since signed contracts with multiple semiconductor companies, including Microchip Technology, where Embedd is enabling Zephyr RTOS support.

The company was founded by Michael Lazarenko (CEO), Maxim Gorinov, and Valentin Gololobov — three Ukrainian tech entrepreneurs who previously ran a hardware company. That earlier company absorbed two successive blows: the chip shortage that ran from 2020 to 2022, which halted production and exposed how fragile tight chip-software coupling is in practice, and Russia's 2022 invasion of Ukraine, which ended the company entirely. Embedd is the response.

The integration overhead

Building a physical AI device — a robot, an autonomous vehicle, a medical device, an industrial system — requires stitching software to hardware at the semiconductor layer. Each chip has its own instruction set, its own peripheral interface specifications, its own configuration registers. Writing the software integration layer that makes a specific chip usable by the application layer above it typically takes engineering teams months per chip. When a chip changes — supply disruption, an upgraded model, a design revision — the integration must be rewritten.

Embedd builds a digital twin of the semiconductor: a software model of the chip's behaviour, interfaces, and capabilities. AI agents then read the chip's technical documentation and auto-generate the integration code that makes the chip usable by the software layer. The company says this reduces integration time by up to 6x, compressing months to weeks.

The immediate addressable market is the physical AI stack: robots, autonomous vehicles, drones, and medical devices are all hardware-intensive and all subject to the same chip-integration overhead. The platform is hardware-agnostic; the Microchip Technology contract demonstrates that chip makers themselves are a customer type, not just device manufacturers.

The Seedcamp thesis

Seedcamp — which has backed Sherpa (Munich, enterprise AI workforce management) and participated in Uncovr (Paris, surgical AI) this year — is here making its third ProYarn-tracked early-stage infrastructure investment in 2026. The common thread is software that handles a technical overhead that engineering teams currently absorb manually: external workforce lifecycle management in Sherpa's case, chip integration automation in Embedd's case.

The round does not disclose a founded year. The company's commercial launch was April 2026, with Lazarenko describing years of hardware experience before the pivot to integration automation.