Aster

What it does

Aster is developing autonomous research systems that aim to automate open-ended scientific discovery. Unlike current systems that rely on benchmarkable goals, Aster's technology is designed to explore novel knowledge without predefined targets. The company has demonstrated results across three benchmarks: Andrej Karpathy's NanoChat, ProteinGym, and the NanoGPT speedrun. They have also scaled autonomous research to handle millions of tokens per second in experimentation.

Who it is for

Aster's technology is intended for researchers and organizations in fields such as materials science, drug discovery, and AI development. By compressing decades of research into hours, it could accelerate innovation for labs, universities, and corporate R&D teams that need to discover novel technologies faster.

Why it matters

Traditional research is slow and often limited by human capacity. Aster's approach could dramatically speed up the discovery of new materials, drugs, and AI architectures. The company emphasizes safety and societal benefit, aiming to deliver tangible benefits while developing the technology responsibly.

Launch signal

Aster is a Y Combinator-backed startup (the first YC neolab) and has published results on multiple benchmarks. They have also shared technical blog posts, including "An LLM Made from Logic Gates" and "Scaling Autonomous Research to Thousands of Agents." The company is currently in seed stage and hiring.

Brand and naming

The name "Aster" suggests a star or a celestial body, evoking ideas of exploration and discovery. The tagline "The First YC Neolab" positions it as a new kind of lab within the YC ecosystem. The brand conveys ambition and a focus on fundamental research automation.