Tutor Intelligence scales US robot deployment and vertical manufacturing
Tutor Intelligence has deployed robots to ship over a quarter of a billion products while building domestic manufacturing capacity in Massachusetts. The company argues that generating real-world data through paid work is the key to training autonomous robot models.
Tutor Intelligence, a 160-person company based primarily in Boston, has deployed robots across the United States to help manufacture and ship more than a quarter of a billion products. The company operates two distinct robot form factors: Cassie, a single-arm robot equipped with suction cups for heavy-duty pallet work, and Sunny, a wheeled semi-humanoid designed for dexterous manipulation. To support this expansion, Tutor manufactures its own hardware in Watertown, Massachusetts, initially sourcing components like actuators and cameras from OEMs but increasingly moving production in-house as volume grows. This verticalization is driven by the view that the existing US supply chain is too small and misaligned for general-purpose robots. CEO Josh Gruenstein noted that the US deployed 38,000 industrial robot work cells in 2025, a figure he described as small compared to global contexts, and argued that the US has a strategic advantage in building new industrial capacity with general-purpose robots, whereas China may face challenges transitioning from its specialized automation infrastructure.
The company’s core strategy involves connecting robots to human operators, or tutors, to generate training data for frontier AI models. This creates an "internet of robot training data" where economic value and data generation are aligned, as customers pay for the work the robots perform. Tutor maintains a 100-robot fleet in Boston called Data Factory One, described as the largest semi-humanoid data collection farm in the US, which is used for real-world reinforcement learning. Robots are priced at approximately $16 per hour, with ongoing inference and teleoperation costs now exceeding upfront capital expenditure. Gruenstein stated that the flow of real-world data from deployed robots is the most valuable long-term asset in robotics, more so than specific model architectures. The company targets small and mid-market businesses, noting that nine out of ten US factories and warehouses currently have no robots. Average timelines range from 40 to 60 days from first call to signed contract, and about 40 days from contract to on-site deployment.