Before founding his current company, Manthan Pawar spent years developing and deploying warehouse robots. One project helped shape his thinking: a robot designed to lift boxes and stack them on pallets. After a year of development, it worked in the lab but failed at a customer’s warehouse. Its suction gripper pulled freshly attached shipping labels off the boxes, causing the boxes to fall. The team had tested it with boxes whose labels were already firmly attached, missing a small but crucial detail of the customer’s operation. For Pawar, the experience showed how much more it takes to make a robot useful in a working warehouse.
Today, that lesson informs his approach at Brooklyn-based Destro AI, which builds software to coordinate warehouse robots, employees, and operational systems. It is the latest chapter in an interest that began with his university’s robotics club, where he helped build a badminton-playing robot from scrap. He later studied robotics at NYU Tandon, working on a device for lymphedema rehabilitation and a sensor-equipped Parkinson’s rehabilitation brace. Roles at HAI Robotics, Dexterity, and Levata then brought him into warehouse automation, where he learned how engineering, deployment, and customer needs determine whether a promising robot becomes a useful product.
The opportunity behind Destro became clearer when Pawar helped build a robotics systems-integration business unit. Evaluating suppliers put him closer to the buyer’s perspective: which technology could solve a particular warehouse problem, and what would it take to make that technology work? He encountered capable hardware at increasingly accessible prices, yet adoption remained difficult. His conclusion was that manufacturing a reliable robot and enabling it to operate within a complex workflow require distinct expertise. Hardware companies must solve assembly, sourcing, quality, and cost. Warehouse operators also need software that understands changing priorities, assigns work, and coordinates machines with the people around them.
Destro addresses that coordination through two complementary systems. MothershipOS manages work across the operation, deciding which robot or worker should perform a task, where it should happen, and when. VisionOS brings intelligence onto individual robots, supporting perception and physical interaction through models trained on human demonstrations. Pawar describes reinforcement learning at the core of MothershipOS, working alongside VisionOS’s robot-level control. The distinction matters because moving a cart or handling a pallet is only one part of a larger sequence. A useful system must also determine which load needs attention next and how that movement fits the surrounding work.
Yusen Logistics has provided a production setting for testing that approach. Its collaboration with Destro initially focuses on cart movement within transload operations, where freight passes through receiving, staging, and outbound shipping. Destro’s platform analyzes warehouse conditions and coordinates assignments across employees and autonomous mobile robots. Pawar describes the deployment as a process of pressure-testing assumptions under conditions where reliability has immediate consequences. Keeping an operation running with minimal intervention requires attention to details that a controlled demonstration can overlook. The partnership also identifies pallet movement and AI-powered workflow verification as areas for future evaluation.
The commercial model follows the same operational focus. Destro offers a robotics-as-a-service solution, with Pawar describing customers as buying throughput capacity. He estimates warehouse labor at roughly $26 to $36 per hour, compared with approximately $4 to $5 per hour for Destro’s solution. Those figures are his broad comparison rather than a detailed accounting of a specific deployment, but they illustrate the economics he wants customers to evaluate: the cost of completing repetitive material movement. Tasks such as pushing heavily loaded carts across a facility can consume substantial time and physical effort, making them a practical starting point for automation.
Working across different robot manufacturers introduces its own constraints. Pawar acknowledges a fragmented market in which access to hardware and control interfaces varies. Destro’s approach combines integration with robot-level intelligence, allowing MothershipOS and VisionOS to communicate as work unfolds. Its hardware-independent ambition therefore depends on compatible platforms and manufacturers willing to provide the necessary access. Pawar sees a developing division of labor: hardware specialists improve the machines, while software companies build the intelligence that makes those machines effective within real operations.
Destro announced an $8 million seed round in September 2026, co-led by Base10 Partners and Bonfire Ventures, with participation from CoFound Partners. The company says it is generating commercial revenue and operating in production, with the funding supporting deployments and engineering growth. For Pawar, the underlying test remains rooted in the lesson of the failed palletizing robot. A machine can perform exactly as designed and still miss the conditions that determine its usefulness. Destro’s task is to build intelligence around those conditions, connecting individual robotic capabilities to the demands of the warehouse as a whole.
