Proteus gains language controls expands in Europe

Proteus now understands natural language and turns spoken requests into tasks.

At Amazon’s Delivering the Future event in London, the company unveiled the next generation of its Proteus autonomous mobile robot, or AMR, highlighting a key shift in practice for warehouse automation. Testing shows workers can tell Proteus what to do without any programming experience or technical commands, and the robot will figure out the priority, the route, and the timing. “You tell it what needs to be done. It becomes your assistant for material movement,” explained Scott Dresser, vice president of Amazon Robotics. The breakthrough is not just a new feature, but a capability that broadens what a single AMR can handle in a busy fulfillment environment.

Documentation indicates Proteus can move carts weighing nearly 400 kilograms, a heavy-duty payload that aligns with Amazon’s aim of removing strenuous tasks from human workers. The Proteus platform is already deployed at 24 U.S. e-commerce fulfillment centers, where it has been used to transport inventory between workstations, transfer containers arriving at docks, and support movement at delivery sites. Beyond dock areas, Amazon says Proteus can navigate to various zones in a facility, enabling more flexible material handling without reprogramming for each new task. The combination of power and adaptability is central to the company’s argument that robots can function as reliable co-workers as scale grows.

The company reports the updated Proteus system is currently being piloted in its laboratories, with European deployment planned for the first half of next year. The Europe push is part of a broader expansion that includes Vulcan, Amazon’s larger, higher-throughput picking robot, and STARK, which handles collaborative tote workflows. Amazon says Vulcan and STARK deployments will spread across Europe in parallel with Proteus, underscoring an integrated automation strategy that aims to coordinate multiple robot types in the same facilities. The move signals a deliberate transition from isolated lab demonstrations toward production-level, multi-robot orchestration in real-world operations.

From a practitioner’s perspective, the shift to natural-language commands is the kind of capability change that changes feasibility in a warehouse setting. On the one hand, letting workers instruct Proteus without special commands lowers the barrier to adoption and expands the set of tasks a single AMR can handle. On the other hand, it raises questions about reliability and safety in noisy environments. Expect tight coupling with warehouse management systems and task queues to be essential; misinterpreted priorities or ambiguous requests could disrupt flows if not mitigated with robust fallback policies and human-in-the-loop checks. In practice, operators will want clear visibility into Proteus’ decisions and the ability to pause or re-prioritize tasks quickly when exceptions occur.

Another practical constraint to watch is the coordination between Proteus and other robots like Vulcan and STARK. While the combined deployment promises more complete coverage of a facility, it also introduces scheduling complexity and potential bottlenecks if multiple robots contend for the same workpieces or routes. The intelligent allocation of tasks across the fleet will matter as Europe-wide deployments begin; latency, error rates, and the need for containment measures around moving machinery will test the limits of on-site safety protocols.

Looking ahead, the European roll-out will be a crucial proving ground for how well language-driven control scales outside the United States and across different facility layouts and regulatory environments. If Proteus can consistently interpret natural-language requests, maintain safe operation with heavy payloads, and synchronize with Vulcan and STARK in diverse warehouses, Amazon will have a stronger signal that wearable-onboarding and adjustable routines can deliver tangible efficiency gains at scale.