First-party human demonstration data

Egocentric Data for Physical AI.

License first-party household and workplace video for robotics, or commission synchronized ego/exo collection to your specification.

Fine Tool Use and Object TransformationDual-tool sequence · Fine motor control · Visible state change

Rights-approved public excerpt

Growing catalogue111 videos make up the documented 10-hour evaluation subset, not the full household inventory6 viewpointsActor POV, dual wrist, and external workplace views800K+Contributor network across 2 collection marketsFirst-partyLicensed GIG Rewards collection programs with telco partners

Two Ways to Start.

Use current inventory when it fits. Commission only the tasks, views, and environments your model is missing.

Collect the Missing Behaviors.

Turn the model objective into a scoped pilot with defined contributors, capture, acceptance criteria, rights, and delivery.

Task
Behavior, setting, and contributor criteria
Capture
Ego, exo, synchronized views, and sensors
Acceptance
Quality, rights, schema, and loader fit
How Custom Collection Works

Inspect Real Data Before You Buy.

Compare household POV, workplace multi-view, and dual-wrist capture using privacy-reviewed public evidence.

Both Wrists Close to the Work.

Two wrist views preserve near-field manipulation while frontal and cropped workspace exo views retain context from the same work event.

Ego
right-wrist ego, left-wrist ego
Exo
frontal, wide whole-body
Alignment
Visually aligned within an estimated 500 ms tolerance; not calibration-grade sensor fusion
Privacy review
360 frames
View All Sample Data

Not a Video Dump. A Release Your Team Can Test.

One accountable path connects the model objective to documented, versioned data and loader acceptance.

  1. Specify

    Define the task, contributor, environment, viewpoint, signals, and success criteria.

  2. Pilot

    Test capture quality, coverage, privacy, timing, and technical fit before scale.

  3. Prepare

    Segment, align, annotate, review, and preserve provenance around the learning signal.

  4. Deliver

    Ship the schema, rights, QA records, checksums, loader, and ingest result.

Built for Due Diligence.

Collection reach, a verifiable operator, and delivery controls stay connected to the data.

800K+

contributors in the GIG Rewards network

Philippines + South AfricaNetwork relationships include Smart Communications, Vodacom, and MTN.

Markets
2
Telco networks
7

Network reach supports sourcing. It is not automatic project capacity. Eligibility, consent, timing, and rights are confirmed in the written scope.

10xme Technology Inc. operates both EGXO Data and GIG Rewards, connecting collection, data operations, and contracting.

Consent and eligibility
Connect approved assets to the contributor, environment, notice, consent, and release requirements.
Provenance and versioning
Trace source assets, protocols, transformations, checks, reviewers, and release versions.
Permitted use and transfer
Define training, derived-model, retention, redistribution, security, and deletion requirements in writing.

Start With the Brief.

Tell us what the model must learn. We will identify whether current inventory fits or define the collection required.

Scope Custom Collection