Instead of selling a robot and walking away, Tutor Intelligence sends a crew of remote operators to babysit its machines on a customer's floor, then slowly hands control to software as the system learns the job. The Watertown, Massachusetts company announced a $75 million Series B led by Andreessen Horowitz, bringing its total capital raised to $117 million, according to a release carried by Business Wire. Union Square Ventures and Neo, earlier backers, also participated.
Tutor spun out of MIT's Computer Science and Artificial Intelligence Laboratory and was founded by Josh Gruenstein and Alon Kosowsky-Sachs. Its two robot lines, Cassie and Sonny, are not humanoid: Cassie is a single-arm machine built for case picking and palletizing in big-box warehouse aisles, while Sonny is a dual-arm system designed for the finer picking work needed around e-commerce shelving. Both run on the company's own foundation models, which Tutor improves using human teleoperators who guide the robots through unfamiliar tasks inside live customer facilities before autonomy takes over.
A labor-rental model, not a robot sale
Tutor's commercial pitch mirrors staffing agencies more than equipment vendors: customers pay by the hour or by the pick, sign no long-term contract, and make no upfront capital purchase. Tutor handles installation, maintenance and ongoing software updates itself. The company says implementation can take as little as eight hours at a new site, and that roughly a third of its recent customers were existing clients expanding deployments rather than first-time buyers — a retention signal investors weigh heavily in warehouse automation, a sector littered with pilot programs that never convert to recurring revenue.
The approach puts Tutor in a different lane than humanoid-robot makers chasing general-purpose dexterity, and closer to established warehouse robotics players like Locus Robotics and Dexterity, which also deploy task-specific arms rather than bipedal machines. Tutor's bet is that narrower, purpose-built hardware reaches reliable autonomy faster than humanoids still working out basic balance and grasping.
What the funding unlocks
The new capital is earmarked for three things: training Tutor's robot foundation models on a growing volume of real-world deployment data, manufacturing more Cassie and Sonny units, and expanding the sites where they operate. Tutor says its robots already work across manufacturing and logistics networks handling consumer goods such as cereal, snacks, cosmetics and toys, and the company is targeting more than 1,000 robots under contract by the end of 2027.
That target implies a roughly tenfold expansion from current deployment levels within about 14 months, a pace that will test whether Tutor's teleoperator-assisted training loop can scale without ballooning labor costs of its own. Andreessen Horowitz general partner Raghu Raghuram framed the bet around compounding data: each new deployment site generates training examples that improve the underlying model, which in turn makes the next deployment faster to bring online.