Sereact

We just closed $110M to solve one of the hardest problems in tech: giving robots a brain. Following our Series B, we are rapidly expanding our teams in Stuttgart and across the United States. We aren’t just building another chatbot; we’re building Cortex 2.0, the intelligence that allows robots to see, reason, and act in the physical world. To do that, we need more than just engineers. We need builders, thinkers, and operators who want to have a massive impact on the future of work. We are hiring across all departments in both the EU and US If you want to work on a mission that actually moves things in the real world, we should talk.

We just closed $110M to solve one of the hardest problems in tech: giving robots a brain. Following our Series B, we are rapidly expanding our teams in Stuttgart and across the United States. We aren’t just building another chatbot; we’re building Cortex 2.0, the intelligence that allows robots to see, reason, and act in the physical world. To do that, we need more than just engineers. We need builders, thinkers, and operators who want to have a massive impact on the future of work. We are hiring across all departments in both the EU and US If you want to work on a mission that actually moves things in the real world, we should talk.

Our new Shoebox Gripper is more than just hardware, it’s the proof of Multi-Gripper Generalization. We’ve seen the industry get stuck on specialized hardware, leading to a massive "integration tax" every time a new tool is introduced. At Sereact, we’re proving that intelligence is the real multiplier. With Cortex 2.0, we’re running the exact same VLA model across three different form factors: stationary arms, mobile dual-arm setups, and humanoid robots. We didn't build this to become a hardware company. We built it because our AI needed a hand that could finally keep up with its brain. Check out the blogpost in the comments!

We are expanding the Sereact team in Stuttgart to redefine Physical AI. While we are hiring across the board, we are especially looking for Mechanical Engineers who want to bridge the gap between high-level software and robust robotics. Open positions include: - (Senior) Mechanical Engineer - Electrical Engineers / Automation Technicians - Hardware R&D Lead If you want to be part of the core team shaping the next generation of robotics, take a look. Link in comments. Come build the future with us.

Handelsblatt featured our journey last week. From starting at the University of Stuttgart to this $110M Series B. The common mistake in robotics is staying in the lab for too long. We took the opposite approach: we prioritized real-world deployments from day one. You cannot solve physical AI in a controlled environment. You have to face the mess, the unpredictable lighting, and the mechanical tension of a live production floor. That is where the real learning happens. Shipping early and learning from the floor is exactly how we scaled. This funding is validation of that philosophy. We aren’t building a research project; we are building the global standard for industrial AI. We’ve proven that our model works where it matters most: At scale, in the real world. Thanks Lina Sophie Knees for telling our story! Full article in comments!

$110M Series B led by Headline, with Bullhound Capital, Felix Capital, and daphni joining. Air Street Capital, Creandum, and Point Nine re-invested. Behind us: Cortex 2.0, our world model reasoning, running in 200+ industrial deployments on contact-rich tasks like assembly, kitting, and precision placement. Alongside us: strong partners with Syncore, AWL, HÖRMANN Intralogistics, Kardex, Körber and customers such as Daimler Truck AG, Mercedes-Benz AG, BMW Group, Österreichische Post AG, bol, Rohlik Group, and Active Ants who bet on us early and shaped the model with every real-world pick. Ahead of us: scaling with existing customers, onboarding new ones, and opening our Boston office to grow our US team. Physical AI has a data problem that you can't solve in a lab. That's why we've spent the last five years building a data flywheel from real production deployments, not simulation. One model. Any robot. Single arm, dual arm, humanoid, or fixed cell. Thank you to our team, our customers, and our investors for the conviction.

Today we're announcing $110M in Series B funding, the launch of Cortex 2, and Sereact's expansion into the United States. Physical AI has a data problem. Everyone is racing to build world models - systems that let robots reason about what will happen before they act. Most of that work is happening in research labs, trained on synthetic data. A world model trained on simulation learns a world that doesn't exist. We took a different bet five years ago. You can't build real robotics AI in a lab. You build it with a data flywheel fed by real deployments - shipping into production, living with the failures, and letting the model learn from what actually happens on the floor. The numbers show it worked. More than 200 Sereact systems are live across Europe. Over one billion real production picks. One human-in-the-loop intervention per 53,000. Cortex 2, launching today, takes that foundation and adds a world model. The robot runs possible actions against a learned model of physics and object behaviour, picks the one most likely to succeed, and updates in real time. It takes Cortex out of the picking bin and into work where contact matters - assembly under tension, kitting, placement where every millimeter counts. We don't build robots. We don't sell services. We ship one thing: the model that runs on any robot. Single arm, dual arm, humanoid, fixed cell - same brain across all of it. To the Sereact team: every pick in production is you. Thank you. To the customers who bet on us early - Daimler Truck, Mercedes-Benz, BMW, Austrian Post, bol., Rohlik Group, Active Ants, MS Direct, DeltiLog, Monta - we owe this moment to you. The round was led by Headline, with Bullhound Capital, Felix Capital, and Daphni joining. Air Street Capital, Creandum, and Point Nine all re-invested. Thank you for the conviction. Next stop: Boston. More soon.

Meet the new Nimbus gripper. Three suction cups, swappable in milliseconds depending on the item. > Higher suction capacity so items don't slip mid-move. Fewer drops, higher UPH. > Pick detection catches failed grips in real time instead of downstream. > Piston force auto-adjusts per item class through Cortex. Gentle on fragile SKUs, firm on heavy ones, no operator input. > Thread-free cup attachment means tool-free swaps. > The vacuum chamber rotates so the hose always faces away from the AutoStore wall. No mid-cycle collisions. > Vacuum and control lines separated into one compact bundle. Fewer hoses, simpler service. > Cable chain with strain relief keeps cables clear of wear paths. > The whole gripper detaches in minutes when maintenance is due. Same gripper concept. Rebuilt from the inside out.

Most robot policies react. They see what's in front of them, pick the next action, and hope it works out. When it doesn't, errors compound. A missed grasp becomes a retry loop becomes a deadlock becomes a human walking over. Cortex 2.0 doesn't react. It plans. The paper is now live on arXiv. Before the robot moves, a world model generates candidate futures in visual latent space. Each gets scored for task progress, risk, and completion likelihood. The robot commits only to the best path forward. See. Plan. Score. Act. We benchmarked against state-of-the-art visuomotor policies across four real-world manipulation tasks. Cortex 2.0 achieved the highest success rates on all four. Zero human interventions across all benchmarks. One architecture. Single-arm, dual-arm, humanoid. Real production environments. Full paper linked below.

MODEX has been massive so far. If you haven't stopped by yet, Booth B11311 is where you'll find us. Single-arm picking. Dual-arm returns handling. Sereact Lens. All running live. Come see it for yourself.