For robotics teams

Tell us what your robot needs to learn.

Turn specific robotics data needs into targeted real-world collection Campaigns.

Define the task, environment, capture requirements, population and budget. Taught scopes the Campaign with you, finds the right Teachers and collects the data.

First-person view of hands placing a granola bag on a grocery shelf

Built for humanoid robotics teams, embodied AI researchers and manipulation-model developers.

Don't search for data.
Commission it.

  1. 01

    Define

    Tell us exactly what behavior or manipulation your model needs.

    • Unload industrial dishwasher
    • Clean commercial restroom
    • Replace plumbing fixture
    • Stock grocery shelf
    • Operate espresso machine
    • Use specific hand tool
  2. 02

    Target

    Reach Teachers by the traits that matter to your task. Teachers opt in to every attribute they share.

  3. 03

    Specify

    • Capture perspective
    • Lesson duration
    • Task requirements
    • Voice annotations
    • Environment
    • Required objects
    • Number of Lessons
    • Geography
    • Deadline
    • Exclusivity
  4. 04

    Budget

    Set the Campaign budget. It funds Teacher incentives, and incentives drive collection volume.

    BudgetTeacher incentivesCollection volume
  5. 05 · 06 · 07

    Collect, QA, Deliver

    CollectTaught distributes the Campaign to matching Teachers.
    QAEvery submission is reviewed against the Campaign's requirements.
    DeliverYou receive the resulting exclusive dataset according to the Campaign's terms.

You define the curriculum.
Taught finds the Teachers.

A Teacher with a chest-mounted phone repotting a plant

A Campaign, in one screen.

Product concept with example values. Not a live Campaign — Campaigns are scoped with our team.
Example Campaign

Residential Dishwasher Loading

Exclusive Campaign
Need
2,000
accepted Lessons
Target environment
Residential kitchens
Dishwasher types
  • Top-control
  • Front-control
Lesson duration
15–25 min
Capture
  • Head-mounted
  • Landscape
  • 60fps
Teacher targeting
  • Dishwasher available
  • Residential home
  • US
  • Kitchen skill
Requirements
  1. Load mixed dishes
  2. Place utensils
  3. Add detergent
  4. Start machine
Teacher incentivesFunded from budget
Collection volumeScoped with Taught
ExclusivityExclusive to you

Egocentric video data

First-person data you can actually train on.

  • First-person videoReal household and workplace tasks, filmed from head- or chest-mounted phones with the hands and the task in view.
  • Motion dataDevice motion recorded alongside the video — captured only while a Lesson is being recorded.
  • AnnotationsTeacher voice and text notes placed on the timeline, task tags and the accepted trim, in original-recording time.
  • Human-reviewedEvery accepted recording is reviewed against the Lesson's requirements before it can be licensed.
  • Consent and rightsRecorded by consenting US adults under versioned consent records, licensed for commercial model training under a written agreement.
  • Exclusive or sharedLicense from the non-exclusive catalog, or commission a Campaign whose data is licensed exclusively to you.

Two kinds of data. One clear line between them.

Marketplace Lessons

Non-exclusive. Licensed across robotics customers.

Consumer marketplace Lessons are non-exclusive and can be licensed to multiple qualified buyers.

Customer-funded Campaigns

Exclusive. Collected for one buyer.

Campaign dataset
Licensed exclusively to the purchasing customer under the agreed license.
Teacher
Receives a guaranteed one-time award in place of royalties. Exclusivity is disclosed before they record.

What does your robot need to learn?

Prefer email?

data@taught.life
What scale are you considering? (optional)