Aiden Bai

Bai created Million.js while in high school, then continued building tools around React’s behavior and performance. His personal account says he attended the University of Washington for three months before leaving to join Y Combinator’s Winter2024 batch. He now describes work on datasets that teach models to write frontend code. The company’s YC profile frames the broader aim as better verification for coding agents, connecting his earlier performance tools to the problem of checking generated software.[1][2]

Finding the code behind what a user sees

React Grab lets a developer select an element in a running interface and copy context for a coding agent. That context includes the element, its component stack and source locations. It removes an identification step from visual feedback: instead of describing a button and asking the model to search for it, the developer can point to its implementation. Its README includes framework-specific development setup and primitives for building a custom picker.[4]

Understanding and reducing rendering work

React Scan highlights components that render inefficiently in a running application, making performance problems visible near the interface they affect. Its current README also recommends React Doctor for agent-oriented static checks. Million.js operates at a different layer: its compiler and optimized virtual-DOM representation reduce the work of comparing UI changes. The tools therefore answer different questions: which component is doing work, and how much reconciliation work the application needs to perform.[5][6]