Ambi Robotics Inc. is leaning on years of production-hardened operational experience in warehouse robotics by giving third-party robots the chance to license the essential skills needed to perform tasks such as item inspection, picking and placement.

The company’s new AI Skill Suite, introduced today, is a dramatic expansion of its operating system for physical artificial intelligence, paving the way for more intelligent robotics that can be deployed in multiple industries.

Ambi Robotics is best known for its fleet of warehouse robots, which can perform an array of tasks. For instance, its AmbiSort A-Series robot arms are used to sort parcels such as boxes, flats and polybags and separate them into individual mail sacks, while the B-Series is designed for middle-mile sortation and automated induction. It can sort parcels from bulk bins into larger “Gaylord” containers or onto conveyors for further processing.

The company has also built an AI-enabled robotic stacking system called AmbiStack that automates the packing of items onto pallets or in containers. Meanwhile, AmbiKit is a multi-robot kitting system that’s able to pick various items from bulk stores to create custom bundles that can be shipped to customers.

All of these robots are powered by AmbiOS, which is a proprietary operating system that acts as the “brains” of its hardware. AmbiOS is what provides the intelligence for its robots, giving them the ability to handle items with precision dexterity and rapidly adapt to new items, including those it has never come across before.

Like all good AI systems, AmbiOS learns from its previous experiences and teaches itself to improve over time, so that its robots only get better at the tasks they’re assigned to do. So if a robot comes across an odd-shaped package it has never seen before, it will quickly work out the most efficient way to pick up that item and pack it into a container or stack it on a palette. Once it has it figured out, it will “remember” that, and pass it on to all of the other robots powered by AmbiOS.

This rich experience, derived from over 250,000 production hours and more than 150 million consumer packages processed so far, is what forms the basis of the new AI Skill Suite. By isolating the hardware complexities, the AI Skill Suite makes it possible to transfer the knowledge of Ambi’s robots to any third-party warehouse robot system, no matter what its form factor is.

Ambi Robotics Chief Technology Officer Jeff Mahler says these skills are worth their weight in gold, because adaptation in real-world settings is one of the hardest challenges to overcome in the robotics industry. He explained that his company has developed the skills for robots to inspect items, understand what they are, pick them up without damaging them, and then precisely place them where they’re needed, from hundreds of hours of real-world production work.

“We’re making our production-hardened intelligence available in standardized applications that allow select partners to add perception, reasoning and physical world manipulation to diverse robotic hardware configurations,” he explained.

The company is offering four AI skills at launch: Item Intelligence, Inspection, Dextrous Picking and Precision Placement, which can be instantly uploaded to any third-party warehouse robot. Each of the skills is powered by Ambi Robotic’s vertically integrated PRIME-1 foundation model, which leverages advanced 3D reasoning to operate safely and effectively in physical environments.

For Ambi Robotics, it’s a move that makes a lot of sense. Not only does it mean additional revenue from licensing AmbiOS to third-party robots, but it will also allow it to generate even more volumes of data that can be recycled to further improve the underlying PRIME-1 model.

“Our systems have obtained data across an enterprise ecosystem that’s responsible for moving 90% of U.S. parcel volume and 95% of active commercial SKUs to every ZIP code in the nation,” said Chief Executive Jim Liefer. “By launching the AI Skill Suite, we’re allowing our partners to leverage this unrivaled repository of real-world operational data to accelerate the deployment of reliable and commercial-scale physical AI.”