Skip to content

01 · Teach It

Teaching robots real work.

Data, models and simulation

Robots learn from data, simulation and people who know the job. That creates businesses in data collection, simulation, task packages and the rights that let an expert's technique be licensed, reused and paid for.

What we look for

  • A narrow task a customer pays to teach today, at a site that uses the result.
  • Clear rights to the data, demonstrations and models the product depends on.
  • Reuse: what is taught once deploys across sites or robot models.

Questions we ask

  • Who owns the data, and who gets paid when it is reused?
  • Does a skill learned in simulation hold up on a real site?
  • What stops a robot maker or an open model from giving this away?

What gets built

  • Training data for physical tasks
  • Simulation and test worlds for robot learning
  • Task packages and data-rights clearing

Recent reading

Our first essays on Teach It are in review. Subscribe to The Progress Bar to get them when they publish.

This month in Teach It

Partner-approved signals for this stage will appear here, with sources and dates, once The Progress Bar launches.