Maven Robotics has emerged from stealth with $100 million in funding and a deliberately narrow starting point: mixed palletizing in distribution centres. The startup’s wheeled, two-armed robots rearrange boxes from factory pallets into mixed loads tailored for individual stores.
Investors include RoboStrategy, LocalGlobe, Vine Ventures and XTX Markets Ventures. Maven plans to build 250 third-generation robots and begin designing a fourth-generation platform, moving from a handful of customer machines toward a much larger commercial deployment.
A Customer Problem Came Before the Robot
CEO and co-founder Hamza Derbas said Maven had little more than a robot concept and a team in 2024 when it learned that a large consumer-goods company was meeting four rival robotics suppliers. He persuaded the company to let Maven inspect its factories and warehouses, then focused the pitch on automating a complete workflow connected to the warehouse-management system.
Maven won the contract despite lacking an existing robot. After two years of work with that company and several other partners, Derbas says as many as eight robots now operate for 16 hours a day with uptime of at least 99%. Those figures are the company’s own account; TechCrunch did not identify the customers or present independently audited performance data.
The current robot travels on wheels at up to 10 miles per hour and uses two arms capable of lifting 30 kilograms. Vacuum grippers pick and arrange boxes. In mixed palletizing, products arriving from different factories are recombined according to changing store demand, work that Derbas said is currently performed by people moving around warehouses and selecting individual items.
Deployment Data Drives the Learning Loop
Maven’s approach depends on information returned by robots in operation. Derbas described pipelines that can bring field data back within minutes or hours, allowing the team to retrain systems, evaluate results, run ablation studies and redeploy updated weights. This creates a learning loop grounded in customer facilities rather than only in a research environment.
The company’s founders bring industrial and financial backgrounds. Derbas worked in automotive engineering, including electric vehicles, and spent nine years in Apple’s special projects group before starting Maven with his brother Khalid, now its chief financial officer. The company also draws on methods developed in autonomous-vehicle programs, where hardware learns from large quantities of real operating data.
Investor Jack Pearson of RoboStrategy argues that Maven’s industrial-systems culture differentiates it from robotics companies organized primarily around research or a preferred technical architecture. That is an investor’s assessment, not proof that Maven’s machines will outperform alternatives across customers or tasks.
General Purpose Arrives One Task at a Time
Maven calls its robots general purpose, but its expansion plan is incremental. The startup estimates palletization as an $80 billion market and wants to master one valuable workflow before moving into material handling, automation and fabrication. Derbas acknowledged that future tasks require manipulation abilities that do not yet exist.
To gather training data, Maven intends to combine its own systems with third-party providers. It has also developed pincer-like gloves that let human operators emulate the gripper form factor the company wants its robots to use. The objective is to capture examples suitable for later machine training while preserving a connection to practical work.
The choice of wheels is part of that pragmatism. Derbas contends that two-legged robots add complexity, unreliability and cost to many industrial workflows. His argument centers on return on investment, although the source provides no purchase price, customer payback period or labor-cost comparison for Maven’s system.
The $100 million round gives Maven the resources to test whether a customer-led strategy can scale from eight reported machines to hundreds. Its early deployment suggests the system can perform a defined job for long shifts. The harder evidence will come as production grows: safety records, maintained uptime, installation economics and proof that learning from one facility transfers to the next.
