Hemlock GTM brain

category

market profile

• text: Robot learning teams need demonstrations that capture physical contact and force as well as first-person video.
• text: Multiple providers collect human demonstrations for physical AI, using a range of camera, teleoperation and tactile capture methods.

personas

• title: Robotics foundation model research lead
  pains:

  • Video and hand pose alone omit how much force a task requires.
  • Needs diverse demonstrations for dexterous, contact-rich manipulation.
• title: Robotics data operations lead
  pains:

  • Must collect and annotate worker demonstrations at scale.
  • Needs synchronized signals and usable sample data for a training pipeline.
• title: Dexterous manipulation engineering lead
  pains:

  • Robots can fail at grip-sensitive tasks even when visual geometry is clear.
  • Needs task-relevant human force examples to test policy improvement.

icp

• text: Physical AI and robotics model developers training manipulation policies from human demonstrations.
• text: Research labs and robot manufacturers with data collection or model training teams, especially those studying force-sensitive tasks.

message pillars

• text: Make human muscle activity useful to machines.
• text: Add force and contact labels to first-person demonstrations.
• text: Capture the way skilled workers perform manual tasks.

objections

• objection: Will force labels actually improve our model?
  response: Review sample data and run an ablation on the buyer's specific manipulation task before scaling.
• objection: Can this data integrate with our robot and training stack?
  response: Inspect synchronization, format, annotation quality and licensing with Hemlock during a sample evaluation.

competitors

• name: 6thSense
  why: Collects tactile, force and visual human demonstration data for dexterous robot training.
• name: Human Archive
  website: https://www.humanarchive.ai/
  why: Supplies egocentric and multimodal physical AI datasets and develops tactile collection methods.
• name: Tacta Systems
  why: Captures production-line demonstrations with sensorized gloves for robotics data.

influencers

• name: Julian Olschwang
  url: https://xitter.gq/j_olschwang
  why: Founder discussing Hemlock's force data and ideal physical AI users.

keywords

• robotics force training data
• egocentric robot demonstration dataset
• physical AI tactile data
• EMG robot training data
• dexterous manipulation datasets
• robot contact force annotation

voice

• Technical and concise
• Visual and demonstration-led
• Ambitious

signals

• text: A robotics lab is training a new dexterous manipulation or physical foundation model.
• text: A team is expanding human demonstration collection or seeking force and contact modalities.

stack

• EMG wearable

pricing

motion

channels

• X
• product demonstrations
• founder-led outreach

confidence

Founders

• name: Julian Olschwang
  title: Founder/CEO
  role: CEO
  bio: Founder at Hemlock. CMU Electrical and Computer Engineering dropout. Previously built GANs for military robotic data generation at the UCLA Biomechatronics Lab.