Praxis Robotics GTM brain

Praxis Robotics sources real-world human work data from business sites for robotics and physical AI teams.

category

Robotics training data

market profile

  • text: Real-world physical interaction data is costly to collect, and robotics developers use human demonstrations alongside robot and simulated data.
  • text: Praxis connects businesses with sites and workers to frontier labs and humanoid developers seeking egocentric and multimodal data.

personas

  • title: Head of data collection at a humanoid robotics company
    pains:
    • Securing varied real work sites and repeatable capture at sufficient volume
    • Obtaining demonstrations that match the robot team's task and sensor specification
  • title: Embodied AI research lead
    pains:
    • Generalizing beyond controlled labs and synthetic scenes
    • Finding task-specific first-person demonstrations for pretraining
  • title: Operations or innovation executive at a site partner
    pains:
    • Turning specialized work environments into licensable data without interrupting operations
    • Determining worker consent, privacy and data rights

icp

  • text: Frontier AI labs and humanoid robot developers building physical-world models and needing access to real human demonstrations.
  • text: Data-supply partners include large businesses with workers and repeatable physical tasks in factories, warehouses and hospitality settings.

message pillars

  • text: Position human know-how as the foundation of more capable machines.

objections

  • objection: Will footage translate to our robot's training requirements?
    response: Agree task coverage, sensors, metadata and acceptance criteria before evaluating a sample.
  • objection: Will capture compromise worker privacy or site operations?
    response: Review consent, data handling, site rules and capture schedules with the partner before deployment.
  • objection: Can the operation scale beyond a custom one-off project?
    response: Request evidence of repeatable site access, quality controls and delivery capacity in a bounded pilot.

competitors

  • name: EgoScale
    why: Supplies real-world egocentric demonstrations for embodied AI models.
  • name: Nferent AI
    website: https://nferent.ai/
    why: Captures and labels multimodal human skill datasets in industrial sites.
  • name: Unidata
    website: https://unidata.pro/blog/egocentric-data-collection-for-robot-training/
    why: Offers egocentric robotics data collection and enrichment.

influencers

  • name: Dev Karpe
    url: https://www.linkedin.com/in/dev-karpe
    why: Praxis cofounder discussing data samples and business partners.

keywords

  • robotics training data
  • egocentric video dataset
  • human demonstration data
  • physical AI data collection
  • humanoid robot training data
  • vision language action training data

voice

  • Aspirational
  • Human-centered

signals

  • text: A humanoid or VLA team is expanding training beyond controlled lab demonstrations into real working environments.
  • text: A business operates many repeatable physical workflows and is exploring data licensing.

pricing

model: unknown

motion

sales_led

channels

  • LinkedIn

confidence

low

Founders

  • name: Rohan Seelamsetty
    title: Founder/CEO
    role: CEO
    bio: CEO @ Praxis. Forged across factory floors, campaign trails, and niche forums. Ex Computation and Cognition at Penn, Apollo Fellow at Oxford, and Indian Debate Team (#1 ranked team worldwide)
  • name: Dev Karpe
    title: Founder/COO
    role: COO
    bio: Founder at Praxis AI, previously at Wharton's Huntsman Program. Background in investment banking and supply chain ops at a F500 CEO's office.
  • name: Tommy Li
    title: Founder/CTO
    role: CTO
    bio: Founder at Praxis AI. Prev. tech private equity at Morgan Stanley. Industrial tech venture capital at Andreessen Horowitz. Published IEEE researcher at NASA JPL and PerceptIn Robotics. BSE CS and Finance from UPenn.

Back to YC companies