In 2024, Maven Robotics was brand new and didn’t have any “comics of robots and teams of people,” CEO and co-founder Hamza Derbas told TechCrunch.
Still, I heard that a major consumer goods company with logistics needs is in town to meet with four rival robot companies about automation. Mr. Darvas personally entered meetings with companies and asked to visit factories and warehouses in exchange for sharing his views on robots.
“We looked at how people work and focused on flows that could bring immediate value,” he told TechCrunch. “We have shown that our approach to robotics is different. We are not trying to solve a single robot problem. We are trying to autonomously handle tasks end-to-end. The robots are connected to the warehouse management system on one side, and the products are loaded onto trucks on the other.”
They won the deal by beating companies with existing robots. After two years working with that company and several other partners, Darvas says Maven has as many as eight robots working 16 hours a day, with an availability rate of more than 99%.
Now, the startup is emerging from stealth with $100 million in funding from RoboStrategy, LocalGlobe, Vine Ventures, and XTX Markets Ventures, and plans to build 250 third-generation robots and begin designing on a fourth-generation platform.
Their robot sits on a wheeled pedestal, can move at 16 miles per hour, and has two arms that can lift up to 30 kilograms. Their main job is “mixed palletizing”. Wooden pallets loaded with boxed products are transported from various factories to distribution centers, where robots create new pallets with mixed products to send to stores.
“Within 48 hours of putting the product on the shelf, they want to change the combination based on real-time demand. That retailer has an order with different combinations (products). Assemble it. Today it’s all done with human labor, running around the warehouse, picking one of this, one of that.”
At Maven’s Santa Clara facility, the robots work smoothly within the training area, using vacuum suction cups to pick up and line up boxes at just the right speed. A live video screen shows two robots working in a customer facility and an employee walking around them.
Darvas has spent his career in automotive engineering, with a focus on EVs, but before founding Maven, he spent nine years at Apple in the company’s special projects group, which is widely believed to have been developing self-driving cars until it was disbanded in 2024, although he did not mention it. At that time, Darvas and his brother Khalid founded Maven. Mr. Khalid is the company’s CFO after a career in private equity.
Like other physical AI companies, the company relies on veterans of self-driving car efforts who have developed the most sophisticated approaches to training autonomous hardware from real-world data. This requires a data pipeline that returns information from a working robot within minutes or hours. “We then perform retraining, evaluation, and ablation studies to identify the appropriate set of weights, redeploy, and run the loop again.”
In a world crowded with robotics companies, Maven sets itself apart by focusing on the realities of industrial operations. Jack Pearson, an investor in RoboStrategy who backed the company, said the company is unique because of its industrial systems background, rather than a research culture optimized for learning or focused on specific architectures.
Agility, a robotics company going public this fall in a $2.5 billion SPAC deal, may be the most similarly positioned company on the market with its focus on safety and specific industrial workflows. But the company’s robots stand on two legs, and Darvas, while emphasizing his respect for the company, says, “What they’re doing doesn’t make any sense… It’s very complex, unreliable, and adds unnecessary cost. The name of the game here is ROI.”
Understanding the realities of overheated facilities and the needs of the people who run them has helped the startup succeed, but if Maven wants to expand beyond its current workflow, it will need to find some kind of research culture. Palletization could be an $80 billion market, but the next set of tasks the company is targeting will require robotic handling capabilities that don’t yet exist.
Maven’s next big move is to collect more data, train robots to handle materials, and move into automation and manufacturing. Leveraging its own system and leveraging third-party providers, the company has also developed a pair of pliers-like gloves that allow humans to emulate the form factor they want in a gripper. The company calls itself a maker of general-purpose robots, but its strategy is to move toward that goal task by task.
“We are based on solving one customer problem at a time,” Darvas says. “If you focus on solving problems and choose fairly large problems, each problem is a multibillion-dollar market. Then you have a wealth of data to learn these skills.”
That process could be the most viable path to bringing robots into the workplace, or the opportunity to get one shot at the next powerful physical AI model rolling out of a cutting-edge lab.
“We’re not in a race for models; we’re in a race to solve industrial labor and enable this labor at the scale the world needs,” Darvas told TechCrunch.
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
