We are a small, cracked, seed-stage team baking the future of search in San Francisco. We built best-in-class retrieval models that understand text, images, and beyond in every language. Our open-source models already rival industry leaders, power lightning-fast vector search, and have millions of downloads on Hugging Face each month. Backed by advisors from OpenAI, Scale AI, and Google DeepMind, we move fast, ship research to production, and obsess over real-world impact.
Everyone bakes. Engineers ship to production daily. Researchers' models reach thousands weekly.
We do one thing exceptionally well: make unstructured information understandable to AI and to humans.
We don't just build AI—we are AI. Agents automate our workflows, LLMs write our docs, and optimize our operations.
No silos. Researchers sit next to engineers sit next to designers. Magic happens.
Good enough isn't. Every API, model, and interaction must be world-class.
We do what we say. To customers, teammates, and ourselves.