Hebbian Robotics GTM brain

Hebbian Robotics builds open-source data quality infrastructure for robotics and physical AI teams training models on recorded data.

category

Physical AI data infrastructure

market profile

  • text: Physical AI model teams must assess and curate large collections of robot and human demonstration recordings before training.
  • text: HFlow targets quality checking and querying physical AI episodes, including teleoperation and policy rollouts.

personas

  • title: Robotics data engineering lead
    pains:
    • Scaling episode ingestion and quality checks
    • Finding useful recordings across fragmented datasets
  • title: Robot learning researcher
    pains:
    • Training on low-quality demonstrations
    • Comparing and curating data before expensive model experiments
  • title: Physical AI data vendor leader
    pains:
    • Proving the quality of collected robot data to model builders
    • Operating search and evaluation infrastructure at scale

icp

  • text: Robotics and embodied AI developers collecting teleoperation, human demonstration or robot rollout data for model training.
  • text: Data vendors or research organizations with large and growing physical AI recording inventories and technical data teams.

message pillars

  • text: Build on open infrastructure for embodied intelligence.
  • text: Improve data quality for robot learning.
  • text: Advance the analysis of robotics training data.

objections

  • objection: Can this fit our existing dataset and tooling?
    response: Evaluate HFlow on a representative sample and inspect format and integration support before adoption.
  • objection: Why use a new tool instead of internal scripts?
    response: Compare reproducibility and time to identify poor recordings on a held-out collection; open source allows technical review.

competitors

  • name: Foxglove
    why: Adjacent robotics data inspection and visualization tooling using MCAP.
  • name: Encord
    why: Data management and curation platform used in physical AI data workflows.

influencers

  • name: Brandon Ong
    url: https://founderland.ai/founder/brandon-ong
    why: Co-founder and CEO active in the physical AI data ecosystem.

keywords

  • robotics training data quality
  • physical AI data infrastructure
  • robot teleoperation data curation
  • robot learning dataset evaluation
  • robot data search
  • HFlow robotics

voice

  • Research-oriented
  • Open source and developer-focused
  • Mission-driven

signals

  • text: Collecting teleoperation demonstrations or scaling robot policy training datasets.
  • text: Hiring robotics data engineers or investing in data quality for embodied models.

stack

  • HFlow
  • MCAP

pricing

model: unknown

motion

community_led

channels

  • GitHub
  • technical writing
  • LinkedIn
  • developer community

confidence

medium

Founders

  • name: Brandon Ong
    title: Co-Founder
    role: Founder
    bio: Brandon is the co-founder of Hebbian Robotics, which builds APIs that help robotics data teams verify data quality for model training. He previously built Gigit AI, a Sequoia-backed AI customer experience startup. His research on AI for low-resource languages topped SEA-HELM, an LLM benchmark co-developed with Percy Liang's research group at Stanford. He briefly pursued a PhD in Robotics at Columbia and NTU before starting Hebbian.
  • name: Kingston Kuan
    title: Co-founder
    role: Founder
    bio: Kingston spent the last few years at Jane Street building systems that scaled to millions of messages per second. He graduated valedictorian from the National University of Singapore (NUS) in Computer Engineering.

Back to YC companies