Post by HELLA Ventures

We visited Critical Mass, the closing party of #Reindustrialize, and got to see some very interesting startups: Peer Robotics, Conduit, and Kyber Labs. The architecture behind robots has been increasingly more shinier than robots themselves. For decades, industrial automation has been deployed as isolated systems: 1️⃣A conveyor 2️⃣A warehouse AGV 3️⃣ A PLC-controlled production line Each of these is optimised locally. The next wave looks a lot different. If you walk a modern factory floor, the bottleneck is no longer any single robot. It is the integration: every new machine, arm, AMR, and sensor speaks its own protocol, and stitching them together has historically cost more than the hardware itself. Three of the companies I saw are each attacking a different part of that tax, and seeing them side by side made the structure obvious. Conduit attacks it at the system level. Their Interoperable Orchestration Platform abstracts heterogeneous robots, machines, and sensors behind one control layer, so the floor behaves like a single coordinated fleet instead of isolated islands. This is essentially a middleware and real-time scheduling problem, and it is where a lot of the durable value sits. You would think this would have been prioritised long ago, but it hasn't been: AMRs and AGVs from different vendors still have no common protocol to talk to each other. Peer Robotics attacks it at the teaching level. Instead of programming the robot, a human demonstrates the task and the robot learns it directly. That collapses the deployment and reprogramming cost that makes traditional automation uneconomical for variable workflows. Kyber Labs attacks it at the physical layer, building dexterous human-like hands for the high-mix, low-volume manipulation that rigid automation has never handled well. Dexterity is arguably the hardest open problem in the whole stack, since contact-rich manipulation does not generalize the way perception has. The field is shifting from making robots that move to making robots that integrate, learn, and manipulate. The economics of automation are being rewritten at the interface layer, not the actuator layer. For Hella Ventures, this rhymes with what is already happening in automotive. The same shift, value moving from individual components to the integration layer that fuses sensors and defines the system in software, is exactly the transition from bundles of independent ECUs to software-defined, sensor-fused vehicles. Same architectural story, different floor. 🚘🤖⚡️ Soham Bakshi #Reindustrialized #Robotics #EmbodiedAI #CorporateVC #Manufacturing #Detroit #vc