Tutor Intelligence Raises $75M in Series B Funding

A warehouse buying robot-hours is also choosing who gets the call when a machine cannot finish a pick. Tutor Intelligence places that responsibility inside its own operation, pairing industrial robots with remote human assistance and model training. Its customer can pay for work while the machinery continues learning how to do more of it.

That arrangement now has a $75M Series B led by Andreessen Horowitz, announced November 2, 2026, with Union Square Ventures and Neo participating. The Watertown, Massachusetts company says the financing brings total capital raised to $117M and will support robot manufacturing, deployment and foundation-model training. Co-founders Josh Gruenstein, CEO, and Alon Kosowsky-Sachs, CTO, are expanding a business that has to deliver useful physical work before every difficult situation has been mastered.

The industrial-AI opportunity becomes quite concrete in that sequence. A customer wants goods moved and orders completed, while Tutor needs the experience generated by operating robots in real facilities. The new capital supports the machines and the organization carrying both obligations, with the cost of human assistance still an important undisclosed part of the economics.

The funding follows a business built around deployment

Tutor's previous financing offers a clear accounting baseline. In its Series A announcement letter, Josh Gruenstein described a $34M round led by Union Square Ventures and co-led by Fundomo, with Neo participating; the company reported $42M in cumulative funding at that point. The company's blog dates that announcement to December 1, 2025. The new $117M total includes earlier capital as well as the Series A and Series B.

The same letter explains why a robotics research idea became a company. Josh Gruenstein joined Alon Kosowsky-Sachs at an MIT lab in 2020, where they became interested in the data required to teach robots useful physical skills. A commercial fleet could generate experience while serving paying customers. Pursuing that idea required hardware, customer relationships and an operating business alongside the research.

Andreessen Horowitz's investment thesis emphasizes this progression from useful deployments toward stronger models. The investor describes a service that can reach smaller and midmarket industrial businesses through usage-based pricing. That is the commercial ambition behind the financing: make an individual deployment worthwhile, then carry what it teaches into the next installation. The Series B announcement leaves valuation and revenue undisclosed, so the operating model can be examined more closely than its financial results.

Cassie and Sonny bring different work into the fleet

Tutor's Cassie robot handles cases and pallets, using a single arm on a mobile base. Its work includes stacking and unstacking pallets and assembling case orders. Mobility allows the machine to move between work areas, giving an operator more flexibility than a permanently fixed installation. Those product descriptions explain the intended workflows; throughput and performance still depend on the application.

Sonny brings two arms to individual-item handling, with e-commerce fulfillment among its initial applications. Tutor describes a machine that navigates existing warehouse aisles, picks items and takes completed orders toward packing. The company also discusses broader manufacturing tasks as development work. A product roadmap should be read separately from the work a customer can already contract for at a particular site.

Human assistance connects the fleet to Tutor's research. On its research page, the company describes remote operators demonstrating tasks, correcting mistakes and evaluating robot behavior. Those interactions supply training information as well as operational support. The person helping a robot recover can contribute to improvements used beyond that single task.

Tutor also operates Data Factory 1, a 100-Sonny research fleet for collecting data and evaluating models. Those machines belong to the research infrastructure, so that count should not be treated as customer installations. Operating a dedicated data factory alongside commercial deployments gives Tutor different places to develop and examine capabilities, while customers continue to judge the work delivered in their own buildings.

Usage pricing leaves a deployment budget to understand

Tutor's published pricing lists palletizing and depalletizing at $14–18 per robot-hour, with a minimum of 80 hours per month. It also lists a $10K implementation fee and an $8K refundable security deposit, plus freight and site-readiness costs. Case and individual-item picking use quoted per-pick pricing. Customers can avoid purchasing the robot outright while still needing to budget for getting the service into production.

That distinction matters to a plant manager evaluating automation. The relevant purchase includes the machine's work, the support behind it and the terms surrounding deployment. Tutor says support, maintenance and model improvements are included in the usage rate. The structure gives the supplier a continuing role in keeping the equipment useful, instead of ending the relationship when hardware reaches the loading dock.

Timing deserves similar care. Tutor's current pricing page describes a median of about 40 days from signing to production, while onsite installation can be much shorter. Combining those stages into a blanket promise of deployment in a day would leave out preparation and coordination. An operations team has to plan around the whole interval before putting a new machine into a production schedule.

In a September interview with the Association for Advancing Automation, Josh Gruenstein discussed the burden created when customers are expected to supervise robots through edge cases. Tutor's answer keeps remote support within its service. For buyers comparing automation approaches, responsibility for those interruptions belongs alongside the quoted rate and the workflow the machine is expected to handle.

Expansion carries the service promise into more buildings

Tutor reports that its robots now work in almost 4x as many facilities as a year earlier. The company also says nearly a third of organizations receiving robots in the preceding 12 months were existing customers installing additional machines. Those are company-reported expansion signals, with a specific denominator: organizations receiving robots during that period. They suggest customers extending deployments, without supplying a complete picture of revenue, utilization or margins.

The company targets more than 1,000 robots under contract by the end of 2027. Contracts and machines already working at customer sites are different measures, and the announcement does not give a current absolute deployed-fleet total. For the manufacturing operation in Watertown, that ambition still creates a practical assignment: build hardware, place it into suitable workflows and sustain the support around each deployment.

The Series B funds those connected jobs. More machines can create more operating experience, while additional sites also bring different layouts, products and exceptions for Tutor's people to handle. Josh Gruenstein and Alon Kosowsky-Sachs are carrying their original data insight into a larger service operation, where a customer adding another robot creates work for manufacturing, deployment and the remote team waiting for the next difficult pick.