The Thesis
The modern warehouse is a remarkably complex operation, and much of it still runs on manual labor, spreadsheets and educated guesses. Workers walk the aisles, scan barcodes by hand and reconcile discrepancies against warehouse management systems that are perpetually out of date. Inventory inaccuracy costs the industry billions annually as miscounts ripple into stockouts, mispicks, excess safety stock and lost revenue.
Today, the structural pressures are intensifying. E-commerce now accounts for over 20% of retail and continues to grow, driving SKU counts sharply higher over the past decade and demanding tighter fulfillment accuracy across every node in the supply chain. Warehouse footprints continue to expand rapidly, while labor is becoming simultaneously scarcer and more expensive. Annual turnover at large warehouse facilities exceeds 40%, and wages are rising materially, making it harder than ever to staff and retain the teams required to run manual processes at scale.
At the same time, a technological inflection point is underway. Autonomous systems, once confined to research labs and tightly controlled environments, are now capable of operating reliably in the dynamic, unstructured reality of an active warehouse. Edge compute has enabled on-device intelligence and hardware costs have collapsed, improving customers’ ROI dramatically.
The industry has already been transformed by a generation of automation systems for picking, sorting, carrying and retrieval that made warehouses faster but left operators blind to what was actually on the racks. The focus has now shifted to operational intelligence, and specifically to real-time inventory visibility as the foundational layer of the modern warehouse. The companies building this layer are accumulating a proprietary 3D spatial model of the physical world, continuously updated and enriched with historical inventory records and operational context. That data flywheel is the basis for the next generation of warehouse technology, where inventory data becomes the engine for autonomous reasoning, predictive operations and a fundamentally more intelligent way of running warehouse operations.
Introducing Corvus Robotics
Corvus Robotics (“the Company”) is a drone-based warehouse intelligence platform that combines proprietary hardware, software and an AI-first perception stack to give enterprises real-time visibility into their physical inventory. The Company was founded by Jackie Wu, who holds an MS in Robotics from Northwestern, and Kabir Mohammed, who was a founding contributor to PX4, the open-source autopilot system that powers much of the modern drone ecosystem and now serves as Chief Executive Officer, with Jackie continuing to serve on Corvus’s board. Headquartered in Mountain View, California, Corvus has spent years building and refining fully autonomous drones designed to operate in the demanding, unpredictable conditions of a live warehouse.
The Corvus One drone deploys in under a week with no infrastructure requirements (i.e., Wi-Fi, beacons, reflectors, lighting modifications). It operates fully lights-out, including in cold chain environments as cold as -20°F, where Corvus is the only solution capable of operating. Equipped with 14 onboard cameras and a neural network that processes data in real time, the drone flies autonomously from row to row, captures 3D SKU-to-slot mapping, flags discrepancies and delivers real-time intelligence via the Company’s customer-facing software. The system operates around the clock without human intervention, giving warehouse teams accurate, up-to-date inventory data in a fraction of the time manual processes require.
Corvus serves some of the largest warehouse operators in the United States, with deployments across food and beverage, automotive, CPG, pharma, 3PL and other verticals. Its customers deploy the platform to tackle one of the most persistent and costly problems in warehouse operations, knowing with confidence what is where and whether it matches the records. The land-and-expand nature of the business reflects both the measurable ROI customers realize and the depth of the product’s integration into their operations.
The roadmap is focused on extending Corvus’s data capture capabilities to continuously enrich the spatial and operational model it is building of each facility. Trident, which made its debut at Modex earlier this year, attaches to forklifts to capture location data on pallets throughout the facility, addressing one of the most significant blind spots in warehouse data collection. As that model grows more complete and more dynamic, it becomes the foundation for an AI Management System that can inform and ultimately automate the operational decisions that today still depend on human judgment and legacy software.
“Corvus Robotics was founded on the conviction that the physical world generates enormous amounts of data and almost none of it gets captured or acted on,” said Kabir Mohammed, Co-Founder and CEO of Corvus. “Our fleet currently analyzes over a million locations every month, giving our customers an unprecedented view into what actually happens inside their facilities.”
At Catalyst, we have been increasingly focused on the convergence of physical AI and intelligent systems, looking for companies that leverage purpose-built hardware to capture data from the real world and transform it into operational intelligence. Corvus is a natural expression of that thesis. Kabir and Jackie have spent nearly a decade building technology that works in one of the hardest environments, earning the trust of some of the country’s most sophisticated warehouse operators along the way. The data they are accumulating, and the platform they are building on top of it, positions Corvus to define an entirely new generation of warehouse intelligence.
We are proud and excited to lead the $20 million Series A-2, and to partner with Kabir, Jackie and the entire Corvus team as they continue to scale.