Language
Text models had the internet.
Language models learned from a massive public record of human writing, links, and documents.
The missing data layer for embodied AI
Robots have bodies. They do not have experience.
TRACE records how people actually move, work, and cooperate in real spaces: the synchronized motion, depth, video, and audio data the next generation of robots needs and no one can scrape.
MMT
core unit
Chest- or head-mounted multimodal scene capture.
6-20
body sensors
Millisecond-synced pose and motion fidelity.
7/7
device harvest
Field-tested swarm collection and upload workflow.
Early
research path
Fit-first access for research and pre-revenue development.
The problem
Robots do not have a Common Crawl.
Language models had the internet. Vision models had billions of images. Robots need the missing record of physical human work: movement, tools, timing, contact, space, and cooperation.
Vision
Vision models had images.
Photos and video gave vision systems a broad training substrate for recognizing the world.
Robotics
Robots have no equivalent.
Robots need physical human task data: movement, tools, timing, contact, space, and cooperation.
Why now
Three things became true at once.
The robotics market is moving, the AI playbook is obvious, and the capture hardware is finally cheap enough to scale outside the lab.
Capital
The money arrived.
Humanoid robotics is funded like a platform shift, but hardware alone does not teach robots how people actually work.
Data
The lesson is settled.
Across language and vision, more real-world data beat cleverer architectures again and again.
Sensors
The parts got cheap.
Industrial-grade motion sensors, radios, depth cameras, and batteries are now commodity parts.
The product
One core unit. Motion fidelity you scale.
Every kit is built around the MMT, then configured with a body-worn sensor swarm. The result is synchronized scene context and full-body motion without a studio.
The kit
One core unit. Motion fidelity you scale.
MMT core
V3.5 / V4
Chest- or head-mounted scene sensing: wide RGB, depth, audio, barometric anchor, and motion.
Body swarm
6-20 sensors
Pose and motion fidelity for how a person moves, works, and shares space with others.
Baseline kit
~$200 target
V3.5-class core plus six to eight sensors; higher configurations scale with sensor count and MMT version.
Built hardware
Prototype capture hardware with real sensor, storage, power, and camera connections.
How it works
Ordinary work becomes training data.
The network path is intentionally simple: people capture eligible sessions, TRACE validates the data products, and the WELL becomes a governed corpus for behavior models.
Person
Sensor kit
App
The WELL
Behavior models
Every uploaded hour moves through quality grading, fraud detection, consent-aware capture settings, and contributor accounting before it can enter the WELL.
01
Wear the kit
A light multimodal core pairs with a small body-sensor swarm for pose, motion, audio, and environment capture.
02
Record real work
Contributors capture ordinary sessions in real spaces, not staged lab sets or narrow mocap studio routines.
03
Verify and govern
Sessions pass quality, consent, sync, and metadata checks before they become useful dataset inventory.
04
License the WELL
Researchers and builders access governed data, while contributors retain a stake in the value they create.
The WELL
A governed corpus for robots that work with humans.
The WELL is TRACE's in-the-wild human-task corpus: raw enough to preserve future training value, structured enough to trust, and built around the people who create it.
In-the-wild human task dataMultimodal capture streamsContributor-aligned economicsResearch-friendly access path
Data products
Cooperative task packages.
Manipulation, handoff, human-proximity navigation, and workspace-sharing data for policy training.
Access
Open for research, licensed for deployment.
Research, non-commercial work, and product development can build on the WELL; commercial deployment requires a license.
Built today
The capture stack already exists.
Working boards, firmware, device provisioning, session harvest, and local upload tooling are already moving through field tests. The hard next step is scale.
Why it compounds
01
Growing dataset
02
Research adoption
03
Commercial licensing
04
Contributor incentives
The wearable can be copied. The aligned contributor network, governed corpus, licensing framework, and processing pipeline are much harder to recreate once they start reinforcing each other.
Pipeline
Swarm collection dashboard showing completed harvest across seven devices.
Pipeline
Raw capture becomes governed inventory.
The processing path is designed for validation, grading, labeling, accounting, and access control before data is used downstream.
Validation
Governance starts before data earns credit.
The public story should be honest: real-world capture brings privacy, consent, and quality risk. TRACE handles that through capture modes, contributor guidance, and validation gates.
Quality
Only useful sessions count.
Uploaded hours are graded for sync, completeness, signal quality, and eligible task capture before they enter the WELL.
Fraud
Low-effort data is rejected.
Replay, synthetic, duplicated, or non-compliant capture can be filtered before it earns contributor credit.
Privacy
Capture modes matter.
Contributors choose session settings and remain responsible for local consent rules; TRACE supplies controls, guidance, and validation gates.
Who it is for
Contributors create the corpus. Researchers build with it.
TRACE has two primary audiences on the public site. The homepage should make both paths obvious without splitting the story too early.
Contributors
Own a share of the data robots need.
TRACE turns useful daily work into verified training data, with contributors participating in the upside instead of disappearing into the supply chain.
Researchers
Build on data that cannot be scraped.
The WELL gives embodied AI researchers a governed path toward physical-world behavior data, not another synthetic benchmark.
Join the build
Help create the data layer robots cannot scrape.
TRACE is building a practical path from real human work to governed embodied-AI training data. Contributors, researchers, and builders can plug into the system as it scales.