Sereact Raises $110 Million Series B

Sereact, physical AI that works where it matters the most, led by Ralf Gulde, Marc Tuscher, and the team, has raised a $110 million Series B round led by Headline, with participation from Bullhound Capital, Daphni, and Felix Capital. Existing investors Air Street Capital, Creandum (lead of Sereact's 2025 Series A), and Point Nine all returned for this round.

The round supports two strategic priorities: scaling Cortex 2.0, the next generation of Sereact’s robotic brain, and expanding into the United States. A first U.S. office has been established in Boston, with local hiring underway across commercial, application, and engineering roles.

Cortex 2.0 - The robotic brain that anticipates outcomes before execution

Cortex 2.0 introduces a shift from reactive action to pre-action planning. Instead of simply seeing and picking, actions are evaluated before execution.

Built by Sereact, Cortex 2.0 augments a vision-language-action (VLA) model with a world model, generating and scoring multiple future trajectories based on physics, risk, and efficiency. Only the optimal path is executed, with real-time updates as conditions change. Trained on over a billion real-world picks, it brings world models from research into production.

This approach replaces trial-and-error with deliberate planning, enabling high-precision tasks where contact and accuracy are critical, marking the next phase of industrial robotics.

Why it generalizes

Cortex 2.0 plans in visual latent space, capturing transferable patterns of objects, contact, and motion beyond robot-specific kinematics. As a result, the same system from Sereact runs across diverse setups, from single-arm cells to humanoid robots.

Planning compute scales with task cost, more for high-risk tasks, less where recovery is easy. Sereact’s Cortex 2.0 allocates effort where it delivers the most value.

Every pick across all sites feeds back into the model. Cortex 2.0 builds on Sereact’s five-year infrastructure, a closed loop where production robots act as continuous data sources, and a centralized model is retrained and redeployed across the fleet.

Every interaction feeds a continuous learning loop, data is captured, refined, and redeployed, compounding over time and expanding long-tail coverage.

This creates a structural edge: while competitors rely on simulation, Sereact is trained on years of real-world operations at scale.

“You can’t build real robotics AI in a lab. It requires a data flywheel driven by real deployments, operating in production, learning from failures, and improving from real-world outcomes. The results reflect that approach: 200 systems, one billion picks, and one intervention per 53,000, levels unmatched in the industry.”

— Ralf Gulde, CEO and Co-Founder, Sereact

“The system operates in latent space, anticipating how the world responds before action is taken. The focus is not on hardware or services, but on delivering a unified model that runs across any robot, single-arm, dual-arm, humanoid, or fixed systems. As hardware commoditizes, the model remains the differentiator.”

— Marc Tuscher, CTO and Co-Founder, Sereact

Why warehouses first

Warehouses offer unmatched real-world data, high volumes, diverse objects, and strict performance demands, making them the ideal starting point for Sereact.

Cortex learns continuously from live operations across sites, powering over 200 deployed systems and more than one billion picks, with minimal human intervention. This real-world data loop creates a compounding advantage over simulation-based approaches.

“The physical AI opportunity is among the most significant of this generation, with the potential to reshape global supply chains and manufacturing. Ralf Gulde and Marc Tuscher are building toward that opportunity through real deployments, real data, and a model that improves with every pick. Strong customer adoption drives continued expansion, accelerating the data flywheel and reinforcing long-term advantage.”

 — Trevor Neff, Growth Partner, Headline

“After evaluating a wave of humanoid robotics companies, the opportunity stood out in Sereact, an AI operating system designed to retrofit seamlessly into the world’s existing fleet of industrial robots.”

 — Per Roman, Founding Partner, Bullhound Capital

“Sereact represents a new generation European leader, combining industrial expertise with advanced AI and robotics talent to address complex supply chain challenges. At its core lies a compounding data moat, where every warehouse interaction strengthens the system and deepens its advantage. The ambition is to define the category of physical AI for the world’s largest retailers and wholesalers.”

— Antoine Nussenbaum, Co-Founder & Investor, Felix Capital