What iPhone Packaging Reveals About the Robotics Opportunity

In this talk

1. 3:24

   Robotics requires a full-stack business

   Gruenstein says robot intelligence alone cannot solve the problem: companies also need suitable hardware, manufacturing capacity, customers, and products. Tutor’s approach combines those layers and tests its platform through operating business units. He argues that making trade-offs across the stack gives a company more ways to improve cost and capability.

2. 9:11

   Tutor learns from robots doing paid work

   Tutor connects robots with human teleoperators, called tutors, who teach them new tasks and generate physical-world training data. Gruenstein says deployments at paying customer sites ensure that the collected data comes from economically useful work. He also describes combining real-world experience with simulation and internet-scale data, while treating tutor use as a unit economics decision.

3. 17:10

   Cassie and Sunny target factory work

   Tutor designs and manufactures its robots in Watertown, Massachusetts, sourcing components such as actuators and cameras from other manufacturers. Cassie handles heavy work with one arm and suction cups, while Sunny is a wheeled industrial semi-humanoid built for factory conditions; Tutor has 100 Sunnys in its Boston data factory. Gruenstein says the robots address tasks such as moving pallets and packing products, including iPhones, that people still handle by hand.

4. 23:37

   Customer deployments shape the product

   Gruenstein says most U.S. factories and warehouses have never bought a robot, so Tutor pitches a concrete labor cost and avoids putting capital risk on the customer. He reports a typical 40 to 60 days from first call to contract, followed by about 40 days to deployment, with robots often working within hours of arrival. A diverse customer base helps Tutor iterate on a general product instead of building a one-off system for a single large buyer.

5. 33:27

   Manufacturing and data feed iteration

   Tutor updates robot manufacturing processes roughly every two weeks and ships code daily or in major two-week releases. Product, customer success, implementation, and research teams work closely so customer needs can inform changes and research breakthroughs can become new capabilities. Gruenstein also describes a 100-robot Boston fleet for real-world reinforcement learning, and says Tutor plans to keep investing in U.S. robotics supply chain capacity.