Waddle Labs GTM brain

Waddle Labs develops language-model agents that generate and refine robot control programs for robotics developers and researchers.

category

LLM agents for robotics

market profile

  • text: Robot developers face deployment work when adapting control to a new embodiment, environment or task.
  • text: Independent robot model companies compete with large platform labs using different control architectures and data strategies.
  • text: Waddle is early access and its code-generating agent approach differs from end-to-end robot foundation models.

personas

  • title: Robotics software lead
    pains:
    • Writing and tuning control for each new arm, gripper and task consumes engineering time.
    • Need to test generated code and verify outcomes on real hardware.
  • title: Robot learning researcher
    pains:
    • Collecting demonstrations and resetting experiments limits training throughput.
    • Models need adaptation across environments and embodiments.
  • title: Robotics deployment lead
    pains:
    • Non-experts need a way to specify tasks and correct failures.
    • Latency and safe operation limit deployment outside controlled labs.

icp

  • text: Robotics developers and research labs with real robots, cameras and a task needing a new control policy; account size is not specified.
  • text: Early adopters able to evaluate real-world performance, latency and safety before deployment fit the early access offer.

message pillars

  • text: Use LLM agents to control robots.
  • text: Produce a working program that users can iterate on conversationally.
  • text: Design robot intelligence around human collaboration.

objections

  • objection: Will this work reliably on our hardware and task?
    response: Run a bounded evaluation using the actual embodiment, task and operating environment; compare intervention and success rates.
  • objection: Are latency and safety good enough for production?
    response: Benchmark the workflow and safety controls; the company explicitly calls latency a remaining bottleneck.

competitors

  • name: Physical Intelligence
    why: Develops general-purpose robot foundation models across embodiments.
  • name: Skild AI
    why: Offers a cross-platform robot brain as an alternative model approach.
  • name: Google DeepMind
    why: Gemini Robotics is an alternative robotics model research stack.

influencers

  • name: Hanming Ye
    url: https://www.waddlelabs.ai/research/introducing-waddle
    why: Coauthor of the company's technical research and founder.
  • name: Yiding Song
    url: https://www.waddlelabs.ai/research/introducing-waddle
    why: Coauthor of the company's technical research and founder.

keywords

  • LLM robot control
  • robot foundation model
  • language model robotics agents
  • real time robot inference
  • robot task generalization
  • robot control code generation

voice

  • Research-oriented
  • Accessible
  • Human-centered

signals

  • text: A robotics team is adapting an existing robot to new tasks or embodiments and wants faster policy development.
  • text: A lab needs automated data collection or scene resets during robot training.

pricing

model: unknown

motion

unknown

channels

  • research publications
  • GitHub
  • events

confidence

medium

Founders

  • name: Hanming Ye
    title: Founder
    role: Founder
    bio: Interested in how natural intelligence arises. I also like machines such as 3D printers and robots. https://dozenducc.github.io/
  • name: Yiding Song
    title: Founder
    role: Founder
    bio: Interested in systems that learn theories about the world with human-like efficiency.

Back to YC companies