By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
a16zWhat iPhone Packaging Reveals About the Robotics Opportunity
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
At a glance
WHAT IT’S REALLY ABOUT
Tutor Intelligence’s full-stack strategy to scale robots through real deployments
- Josh Gruenstein explains Tutor Intelligence’s thesis: deploy robots into real parts of the physical economy first, then use that deployment to generate the data, models, and supply chain needed to scale robotics.
- He frames robotics as a full-stack systems problem spanning models, hardware, software, supply chain, customers, product, and go-to-market, where progress comes from making trade-offs across layers rather than optimizing one.
- He describes Tutor’s “tutors” as human teleoperators who teach robots tasks, with the goal of building an “internet of robot training data” created as a byproduct of paid customer work.
- He outlines Tutor’s two robot platforms—Cassie for heavy-duty pallet/case handling and Sonny for more dexterous industrial manipulation—plus what they currently can’t do (notably outdoor work) and where future embodiments may go.
- He details commercialization dynamics: most customers are first-time robot buyers, adoption is driven by demonstrated operational value, and ongoing operational costs (including inference/teleoperation and maintenance) increasingly dominate unit economics.
IDEAS WORTH REMEMBERING
5 ideasStart with the job the robot must do, not the model.
He argues that many robotics efforts over-index on models or data mechanics without a clear end-job, while Tutor started by targeting concrete, high-volume supply-chain tasks and then built the data, models, and deployment engine around that.
Robotics is an “ultimate systems” and full-stack business problem.
He describes robotics as requiring coordinated decisions across intelligence, hardware, software, supply chain, customers, product, and go-to-market; excluding any layer collapses the trade-off space needed to make progress.
Tutor’s core flywheel is real-world deployment plus tutor-driven data creation.
Tutor’s approach is to connect robots in real customer sites to human teleoperators (“tutors”) to teach new tasks, turning operational work into training data and creating a flywheel of deployment → data → better capability → more deployment.
Customers care about performance and economics, not whether a robot is being tutored.
He says customers mostly care that outcomes improve (e.g., cheaper goods), not whether a human tutor or an autonomous model is currently in control; that “seamlessness” is a sign the robots are useful.
Hardware must be co-designed for modern robot intelligence, so Tutor builds its own.
He says the company was pushed to design and manufacture its own robots because existing industrial robots are optimized for a different regime (systems integrators, repetitive motions) and not co-designed with frontier models.
WORDS WORTH SAVING
5 quotesRobotics is not actually full stack like model, hardware, software. It's full stack like model, hardware, software, supply chain, customers, product-
— Josh Gruenstein
If you throw out one of the layers of the stack and you're like, "Well, we're not gonna touch that," you've eliminated 90% of your trade-off space to make progress.
— Josh Gruenstein
We call them up and we're saying, "Hey, have you ever considered a robot for 16 bucks an hour?"
— Josh Gruenstein
The operations community is really, really fantastic 'cause they see a thing work and it's like, "I... It works."
— Josh Gruenstein
We have this framework of the value of a robot is you take how much work a robot can do and you divide it by its cost, and that's it.
— Josh Gruenstein
High quality AI-generated summary created from speaker-labeled transcript.
