DeepReach Inc. GTM brain
DeepReach supplies real-world wearable capture data to physical AI developers through a network of local collection partners.
category
Physical AI training datamarket profile
- text: Robotics and world-model developers need demonstrations of real work across varied environments rather than a single controlled setting.
- text: The business has two sides: model builders buying training data and local partners operating capture networks.
personas
- title: Head of data at a robotics foundation model labpains:
- Training data lacks variation in tasks, workers and physical environments.
- Collecting and quality checking real-world demonstrations at scale requires hardware and local access.
- title: Robotics research leadpains:
- Policies trained on narrow settings may not generalize to messy workplaces.
- Needs relevant, usable and permissioned capture data for targeted manipulation tasks.
- title: Local data collection partnerpains:
- Has access to real workers and locations but needs devices, task specifications and reliable payments.
- Must assess variable approved-hour earnings before committing resources.
icp
- text: Frontier model labs and robotics companies training physical AI with substantial real-world data requirements.
- text: Local operators able to access diverse workplaces and manage teams collecting approved task footage.
message pillars
- text: Offer partners a way to start a collection business.
- text: Match collection tasks to each partner's capabilities.
- text: Capture work across many settings for model training.
objections
- objection: Is the data sufficiently diverse, consented and useful for our model?response: Review representative samples, permissions, collection conditions and quality criteria against a targeted evaluation task.
- objection: Can a collection partner count on the advertised earnings?response: Treat the published hourly example as illustrative; actual earnings depend on available tasks, geography and accepted quality.
competitors
- name: XDOFwhy: Collects robotics training demonstrations using teleoperation and wearable sensors.
- name: Scale AIwhy: Provides physical AI collection and annotation for robotics labs.
- name: Build AIwhy: Offers egocentric physical labor datasets and its own capture hardware.
keywords
- physical AI training data
- robotics egocentric video dataset
- wearable robot data collection
- human demonstration data
- robot manipulation training data
- real world multimodal data
voice
- entrepreneurial
- practical and accessible
- community-focused
signals
- text: A robotics lab is expanding physical AI training and seeks data from more settings or task categories.
- text: A prospective local partner has access to workplaces, collectors and enough capacity to manage multiple devices.
stack
- wearable stereo capture devices
hiring
- Robotics Research Intern
- Member of Technical Staff, ML Infrastructure
- Member of Technical Staff, Computer Vision
- Member of Technical Staff, VLA
- Teleoperation Operator
pricing
model: unknown
motion
sales_ledchannels
- YC directory
- partner applications
- SEO
confidence
mediumFounders
- name: Tim Lititle: Founder/CEOrole: CEObio: Tim Li is co-founder and CEO of DeepReach. He has spent a decade building cross-border networks of people: HireIO, a Silicon Valley cross-border staffing platform he scaled to eight-figure ARR, and Talex, an expert network of ~7,000 annotators across ten countries working on LLM training data. At DeepReach he leads strategy, hardware operations, supplier networks and commercial. He works across US and Chinese supply chains and is bilingual in English and Mandarin.
- name: Chris Liutitle: Founder/CTOrole: CTObio: Co-founder at DeepReach Inc. Founding MLE & tech lead of Aven scaling to #1 in DM volume in the US. Research Scientist at Meta working on core ML systems. PhD in computer vision from USC. B.S. from Peking University.