01
Request access
Tell TRACE what work you can capture, where it happens, and which hardware path fits the session.
Contributors
Turn real work into the data robots need.
TRACE is opening an early contributor path for people who can capture useful physical work with lightweight sensor hardware. Verified sessions help build the WELL: governed human task data for embodied AI.
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.
MMT
required core
Chest- or head-mounted scene and task sensing.
6-20
body sensors
Configured by task, motion fidelity, and setting.
~$200
baseline target
V3.5-class core plus six to eight sensors.
Gated
validated hours
Only compliant, validated hours count.
How contribution works
Request access. Capture useful work. Submit sessions for validation.
TRACE is not asking contributors to perform for a camera. The goal is ordinary task data with enough sensor fidelity, consent discipline, and quality control to make it useful.
02
Configure the kit
The MMT records scene context while body sensors capture how you move, reach, turn, lift, and share space.
03
Capture real sessions
Useful sessions come from real homes, shops, kitchens, benches, job sites, and workspaces rather than staged routines.
04
Submit for validation
TRACE checks sync, completeness, quality, capture mode, and fraud signals before a session enters the WELL.
Eligible sessions
The first priority is useful, ordinary physical work.
Strong contributor sessions show movement, tools, timing, proximity, and shared space. TRACE is especially interested in cooperative task data that is hard to synthesize or scrape.
Field hardware
Prototype TRACE capture hardware connected during bench and field testing.
Tool useHandoffsWorkspace sharingKitchen workBench workHuman-proximity navigation
What counts
Ordinary work is the point.
A simple, well-synced session of someone doing real work can be more valuable than a polished clip. The system cares about physical behavior, context, and validation.
Validation
Validation protects contributors and researchers.
Contributor trust depends on clear gates. Sessions have to pass quality, consent-aware capture settings, and fraud checks before they enter the WELL or earn credit.
Quality
The data has to be useful.
Sessions are checked for sensor sync, signal completeness, task relevance, and enough context to train future behavior models.
Consent
Capture rules are part of the work.
Contributors choose capture modes and are responsible for local laws and permissions. TRACE supplies controls and guidance.
Fraud
Credit follows real production.
Synthetic, replayed, duplicated, low-effort, or non-compliant capture can be rejected before it earns contributor accounting credit.
The upside
The people who produce the data should participate in what it earns.
TRACE is built around contributor alignment. Exact terms belong in onboarding, but the principle is simple: useful verified work should not disappear into someone else's dataset.
Contributor share
Validated hours participate in the upside.
TRACE is designed so the people producing the dataset participate when commercial licensing creates value. Exact terms belong in onboarding, not vague marketing copy.
Early network
The first useful hours matter most.
Early contributors help seed the dataset, prove the capture workflow, and establish the task categories future researchers and builders can use.
Start here
Request contributor access for early hardware.
Tell TRACE what you can capture, which settings you work in, and what kit configuration would make those sessions useful.