Tutor Intelligence raises $75 million from a16z to build robots it rents out by the hour

Tutor Intelligence, a Watertown, Massachusetts company whose warehouse robots work with help from remote human operators it calls "tutors," has raised a $75 million Series B led by Andreessen Horowitz, it said on Nov. 2. Union Square Ventures and Neo, which led its Series A and seed rounds, joined, bringing the total raised to $117 million.

Much of the money will go into robots that customers never buy. They pay by the hour or by the pick, make no upfront purchase and sign no long-term contract, while Tutor deploys and maintains the machines and sends them new software. a16z says hourly pricing turns customers' capital spending into operating costs; the capital still has to come from somewhere, and here it comes from Tutor's investors. The round is fleet capital as much as research money. The release says it will pay for training Tutor's robot foundation model and for building and deploying its two robots. Cassie has a single arm for case picking and palletizing; Sonny has two arms for picking individual items, and both run the same model, according to a16z.

The paid work doubles as the training set. Every task the robots complete produces the real-world data robot models lack, Tutor says, and the tutors' teleoperation covers what the model can't yet do. The company says its robots now work in almost four times as many facilities as a year ago and that nearly a third of the organizations that received robots in the past 12 months were existing customers adding more. It is on track for more than 1,000 robots under contract by the end of 2027, all built in the United States. "We can hardly build robots fast enough," said CEO Josh Gruenstein, who co-founded the company with Alon Kosowsky-Sachs out of MIT CSAIL.

The tutors are also the cost to watch. As long as people step in on a large share of tasks, each hour Tutor sells has to pay for a remote operator as well as a robot.

Ultra, a Brooklyn startup that also rents out warehouse robots, takes the other route on both counts. Its customers pay an up-front integration fee and then a monthly support fee, and its AI comes from Physical Intelligence. Tutor asks for nothing up front and trains its own model, so it carries more of the hardware bill and all of the research.