Get a Quote for Egocentric Data Collection

A quote request starts the same way an order does: with a written spec rather than a request for a price list. The spec names the tasks or skills to capture, the scenes and environments, the objects and contributor demographics, the streams and labels needed, the accept criteria a take has to clear, the delivery format, and the timeline. That written spec is what gets quoted — reach, stream count, validated hours, and turnaround all move the number independently of each other.

Once the spec is in, it is typically proven at small scale first with a Spec Pilot before the full quote is confirmed. The pilot surfaces whether the reject rate on the spec is manageable, whether coverage for the requested tasks and regions is thin or deep, and whether the stream count requested is actually necessary — all of which affect the final number more precisely than a pre-pilot estimate can.

A scopeable quote request answers: which tasks or skills, in which environments and regions, performed by which kind of contributors, captured with which streams (video, depth, IMU, hand pose, body pose, segmentation, action labels), and checked against which accept criteria. Target volume — hours, episodes, contributors, or environment variants — is stated in your own terms rather than selected from a fixed package size.

The request should also name a delivery format — RLDS, LeRobot, WebDataset, or whichever structure your training loader already expects — and any timeline constraint. Naming the format and timeline up front means the quote reflects the actual delivery requirement instead of a generic assumption.

The skill spec is the document that converts a general request into a scoped, quotable job: it is what names the tasks, environments, contributors, streams, labels, and accept criteria that the quote is actually priced against. Without a written spec, a request cannot be scoped beyond a rough estimate, because reach, stream count, and turnaround are all spec-dependent rather than fixed.

Because the spec also defines what counts as a validated take, it is the same document that QA and acceptance run against later — the quote, the Spec Pilot, and the delivered batch are all checked against one written spec rather than three separate descriptions of the work.

A quote covers the scope defined in the spec: the Spec Pilot (where used) to prove accept criteria hold on real footage, the full capture and QA pass against those accept criteria, delivery of the synchronized streams and labels in the agreed format, a QA report documenting what was reviewed, and documentation of the consent chain for the footage included. Everything quoted and delivered carries the same perpetual, buyer-owned commercial license.

What a quote does not cover is a fixed unit price — because stream count, validated-hour volume, coverage depth, and turnaround all move the number independently, the quote is a number against your specific spec, not a rate card entry.

There is no list price. The number in a quote is driven by how many synchronized streams each episode needs, how many validated hours the spec requires, how thin or deep existing coverage already is for the requested tasks and environments, and how fast the first batch needs to ship. Billing, once an order is confirmed, is strictly per validated hour: footage that fails QA is rejected and recaptured at the provider’s cost, not billed to the buyer.

See the full breakdown of how each of these levers moves the number on the pricing page. This page covers how to request the quote itself; the pricing page covers what the quote is actually made of.

A Spec Pilot is a small, fast first batch run against the written spec before the full quote is confirmed. It proves whether the accept criteria hold on real footage, whether the reject rate is manageable, and whether the requested reach and stream count are actually necessary — turning the quote from an estimate into a number backed by a real small-scale result.

Running a pilot is optional but is the fastest way to get a confirmed number against a spec rather than relying on an estimate alone, particularly where coverage for the requested tasks or regions is not already deep.

Every quote and every delivered batch carries the same terms: a perpetual, buyer-owned commercial license backed by a documented consent chain for the contributors included, and an anonymization pass that detects, blurs, and has a reviewer confirm the removal of faces, license plates, and screens before delivery. These terms do not vary by order size — they are part of what a quote always covers, not an add-on negotiated separately.

Where coverage for the requested tasks, environments, and regions is already deep, the median time from signed spec to first validated batch is 48 hours. Where coverage is thin — a region, task, or condition combination with little standing contributor coverage — turnaround runs three to four weeks, because recruiting has to happen in parallel across more regions before capture can scale.

A Spec Pilot gives a real, small-scale timeline figure specific to the submitted spec, rather than relying on the general figures above as a stand-in for your particular request.

Once a quote is accepted, the engagement moves to a Spec Pilot (if one has not already run) to confirm the spec and accept criteria on real footage, then scales to the full capture volume. Capture, review, and delivery run in batches rather than one large drop, so validated data and passing QA are visible as the program progresses rather than only at the end.