AI Is Quietly Reshaping Warehouse Robotics, and One Founder Explains How
OSARO co-founder Derik Pridmore says the real breakthroughs in warehouse automation come not from flashy demos, but from continuous learning, hardware flexibility, and safety you can actually trust.
Key points
- OSARO, co-founded in 2015, builds AI software that gives industrial robots better perception and control at warehouse scale.
- CEO Derik Pridmore argues that hardware-agnostic design, meaning software that works with many different robot bodies rather than just one, is more valuable than any single impressive demo.
- The FCC (the U.S. Federal Communications Commission, the regulator that oversees communications and, increasingly, technology imports) recently moved to limit U.S. imports of new humanoid and mobile robots, putting domestic AI policy in the spotlight.
- Pridmore brings over a decade of investment experience, including early bets on DeepMind and Clarifai, two AI research organisations that later became well-known in the industry.
When a package needs picking in a modern warehouse, the decision about which robot arm grabs it, at what angle, with how much force, now often belongs to an AI system running quietly in the background. That shift from rigid, pre-programmed machines to adaptable, learning-based automation is the subject of a recent conversation published by The Robot Report.
What did Derik Pridmore actually say?
Pridmore's central argument is that the biggest gains in robotics no longer come from better hardware alone. They come from smarter software: systems that keep learning from real-world errors, that work across different robot brands, and that are monitored continuously so problems surface before a whole production line stalls.
He describes a journey that many warehouses are only part way through. Early robotic systems had narrow perception, meaning they could only recognise and handle objects they had seen many times before in controlled conditions. Newer AI-driven systems can generalise: they figure out how to handle a box they have never seen, in a position they were never trained on.
But Pridmore is careful not to oversell. Balancing specificity, reliability, and safety, he says, is still where the hard work lives. A robot that is impressive in a demo but unpredictable on a real shop floor is a liability, not an asset.
Why does the FCC ruling matter here?
The episode also covers a significant policy moment. The FCC recently announced limits on U.S. imports of new humanoid and mobile robots, a move that puts a spotlight on how dependent American warehouses and factories might become on foreign-made machines. If companies cannot import certain robots freely, the pressure to build and software-enable domestic alternatives grows fast.
For warehouse operators, this is worth watching. Supplier choices made today could become supply-chain headaches tomorrow if import rules tighten further.
What should ordinary people take from this?
Most of us will not run a warehouse, but we all receive packages from them. Faster, more accurate picking means fewer wrong orders and quicker deliveries. It also means the workforce inside those buildings is changing: some repetitive jobs are being replaced, while demand grows for technicians who can oversee and fix AI-driven systems.
If you work in logistics or operations, the practical signal from Pridmore is clear: solutions that lock you into one hardware vendor are riskier than ones that adapt as your needs change. Real-world monitoring matters more than benchmark numbers.
Watch for warehouses advertising AI-powered fulfilment as a selling point when they partner with retailers. That is where these systems are already at work.
