Perit — frontier training data, RL environments and benchmarks, graded by people who do the work

We collect and provide image training data for AI models to learn from.

Perit is a training-data company. Everything is recorded by the people who actually do the job, labelled by our own team, and licensed to you — including embodied data for robotics.

Get in touch See what we collect

4,000+ Hours of audio delivered 9 locales

We collect and deliver data in these languages:

de_DE German fr_FR French it_IT Italian es_ES Spanish pt_BR Portuguese hi_IN Hindi zh_CN Chinese ko_KR Korean ja_JP Japanese

Our track record

What we've delivered so far.

Speech is our biggest line of work — 9 locales, audio delivered every working day, and a quality report with every batch. The same team, the same checks and the same reporting run our video, image and text work.

Services

Collect it, annotate it, or ask for the set that already exists.

Data collection
Speech delivered at scale. Video, image and text collected to spec.
Speech, video, image and text · nine locales

Data annotation
The Foundry bench — transcription, alignment and labels, any data type.
Audio, image, video and text · one calibration gate

Request data
Name the input your model fails on. We collect it, transcribe or label it, and license the set with the spec and terms published.
Priced by the unit it is delivered in

How buying works

Four steps, and the first one is a phone call.

  1. Tell us the failure
    Not the dataset — the input your model gets wrong. A clip, a photo, a paragraph is enough to start.
  2. Written spec and quote
    Quantity, locales, devices, conditions, annotation depth — and the contributor rate you will be paying, stated as its own line.
  3. Pilot batch
    Five percent of the volume, or ten to thirty episodes for embodied data, delivered first with a quality report. If the spec was wrong, the re-run is on us.
  4. Ongoing delivery
    Drops into your bucket in the region you chose, with the manifest, consent references and reviewer chain on every item.

Physical Intelligence

Embodied data, collected by people who do the task.

Egocentric video, depth, teleoperation and handheld-gripper demonstrations, recorded in real homes, shops and workplaces across nine locales on the same recruiting, consent and QA machinery as our speech work.

Programs running in nine locales · quote in 48 hours

First-person RGB video of real people doing real tasks.

RGB-D, stereo and LiDAR capture for 3D scene understanding.

Human-guided robot demonstrations, one episode at a time.

In-the-wild wrist-gripper demos for grasp generalisation.

What lands in your bucket

This is a delivery, not a description of one.

Who does the work

Operators, not a crowd.

The bench is built from people who worked the floor, the ward and the claims desk — ex-contact-centre QA leads, claims handlers, ward clerks, shopkeepers. They are recruited for the sector or the locale a program needs, not pulled from a general pool.

Two crafts, one bench. Transcribers write what was actually said, fillers and all. Aligners put a start and end time on every word. A rubric author is someone who signs off on these calls at work, and every rubric has to be applicable by a stranger.

Coverage

Crews on every continent, recruited to the locale you need.

Contributors record on their own phones, in their own rooms, in the accent and the light they actually live in. Nine locales across Europe, Asia and the Americas — and anything beyond it is recruited to order against a locale quota.

Research

Scored against the person who does the job.

Record, film, photograph, write

Whatever we collect, we pay the person who made it.

Pick a task on the workspace, do it on your own device in your own space, and get paid once it passes review. The rate is on the task before you start, and it is the same rate whoever the buyer is.

Independence

Independent, and staying that way.

No frontier lab owns a piece of Perit, and none of them sits on our cap table. Your held-out audio does not end up next to a competitor's. Backed by Y Combinator, F26.