About Goodfire
Goodfire is a research company using interpretability to understand, learn from, and design AI systems. Our mission is to build the next generation of safe and powerful AI—not by scaling alone, but by understanding the intelligence we're building.
Scaling has proven powerful, but today's approach is fundamentally limited: we can't meaningfully understand, debug, or shape what models learn. Every engineering discipline has been gated by fundamental science and AI is at that inflection point now.
We're advancing the science of how AI systems actually work. Treating models as black boxes is an unnecessary handicap—we have access to the structures inside them, and understanding those structures lets us steer what models learn, make them safer and more useful, and extract the vast knowledge they contain. Our goal is to make AI that can be understood, debugged, and shaped like software.
We are a team of researchers, engineers and builders shaping the frontier of AI
Our team includes founding members of interpretability efforts at Google DeepMind and OpenAI, professors on leave, and engineers who have built and deployed large-scale ML systems at organizations like OpenAI, Google, and Palantir.
Many of us helped pioneer core research directions in interpretability—from discovering sparse, human-meaningful neural network features using sparse autoencoders, to automated feature interpretation, to extracting knowledge from superhuman models.
Our team
Contact us
Interested in partnering with Goodfire?