About Mithran
Founded in 2026
Mithran was founded by Eran Shir and Bruno Fernandez-Ruiz, a partnership that began at Yahoo nearly two decades ago and has been building AI systems for the real world ever since.
The founders
A physicist turned serial entrepreneur. Eran founded DIMES, the distributed internet-mapping project, and later founded Dapper, acquired by Yahoo in 2010. At Yahoo he ran the global Creative Innovation Center and Israel R&D center. After a chapter as an EIR at Aleph, he co-founded Nexar in 2015 and led it from idea to a category-defining vision AI network, raising more than $170M and putting AI into production on roads around the world.
A veteran builder of large-scale systems. Bruno was a Senior Fellow at Yahoo, where he led personalization, advertising, cloud, and platform systems. Before Yahoo he founded OneSoup and YamiGo, served as an enterprise architect at Fidelity, and co-founded Meridea. He holds an MSc from MIT. As Nexar's CTO, he spent a decade running city-scale vision AI and developed the evidence-based engineering discipline now embedded throughout Mithran.
The journey
We met at Yahoo. Bruno built platforms for hundreds of millions of users. Eran led innovation. Both learned what it takes to make complex systems survive production.
We founded Nexar because we believed AI belonged in the real world. We spent the next decade putting it onto roads, where every input is hostile and every failure is public.
We gave every function at Nexar the ability to build with AI. The ideas came fast. The path to safe production did not. So we built a third way.
The operating system we built solved a problem every enterprise is about to have, not just Nexar's. So we spun it out as its own company to bring it to all of them, with Nexar as our first customer and lighthouse.
Seven months at Nexar
Behind those deployments sat 6,974 security reviews and 1,042 publish gates. The point was not to slow people down. It was to give the company enough control to let more people build.
What we believe
Guardrails are not a tax on AI productivity. They are what makes the productivity land. Nobody frees a workforce to build without a path to production they can trust.
Whoever builds an agent, including us, does not get to grade its own safety. Judgment belongs to an independent adversarial layer in the path.
A tool that accelerates only developers captures a fraction of the opportunity. The enterprise prize is finance, operations, legal, marketing, and engineering building safely on one governed foundation.
Mithran
If your people are already building with AI, we should talk about what happens next.
hello@mithran.ai