Waddle Labs

About Waddle Labs

Waddle Labs operates at the intersection of artificial intelligence and robotics, providing a system that enables AI agents to write control code for robots. Founded in 2025 by Hanming Ye and Yiding "Vincent" Song, the Boston-based company is part of Y Combinator's Summer 2026 batch. The founders, both Harvard students, possess deep research backgrounds relevant to the company's product. Ye authored a paper on robot policies, while Song's research focuses on program induction. The core offering is an API that translates natural-language prompts and camera feeds into editable control programs for robotic hardware. This approach significantly reduces the manual engineering and task-specific data typically required to program robots for new functions. The system functions by layering an AI agent on top of a robot's foundational control model. This agent can perceive the environment through camera feeds, break down a complex task into steps, generate executable code, and then test and revise that code based on the physical outcome. An operator can intervene by asking the agent for revisions rather than initiating a full retraining cycle. Waddle Labs' architecture is designed for adaptability, allowing the same agent system to work with different hardware configurations, such as various robotic arms, grippers, and cameras, without needing new training data for each setup. The agents build a shared library of parametrized skills—like grasping or folding—which can be composed and reused across different tasks, enhancing efficiency with each deployment. The company launched its early-access API in July 2026, targeting entities looking to automate physical tasks and accelerate robotics research and deployment.