Built by someone who
believes in clarity.

I started Crypducate after watching too many people: friends, family, people in my community: lose money in crypto not because they were greedy, but because nobody ever taught them the basics before they jumped in.

Before this I spent years doing exactly what Crypducate is built on: turning messy, complex data into decisions people can actually use. At Publicis I led analytics across more than $50 million in annual media spend, built systems that surfaced over $1 million in savings, and designed AI-assisted workflows before most people in my industry knew what that meant. At Horizon I built the data pipelines from scratch. The common thread in all of it: making something complicated legible to the people who need to act on it.

That is what Sato does for crypto. Not a summary. Not a glossary. A tutor that figures out where you are and builds from there. Named after Satoshi Nakamoto, the person who believed financial tools should belong to everyone. We think education should too.

Most AI products bolt a language model onto a problem and call it done. Xi's entire research career has been about what happens when you do that carelessly, and how to build something that actually knows what it knows.

Xi is a PhD researcher at Johns Hopkins, published at the top venues in machine learning. His work on KBLaM gives language models a structured knowledge base they can query accurately. Featured by Microsoft Research. Over 1,400 stars on GitHub. His research on uncertainty quantification produces AI systems that tell you when they are not sure, rather than confidently getting it wrong.

That matters for Sato. A crypto tutor that hallucinates answers is worse than useless. Xi's research shapes how Sato is built: grounded in a real curriculum, calibrated about what it knows, and designed to be trustworthy rather than just fluent.

Michael is the person who turns architecture into a working product. At Dealmaker he designed agentic AI harnesses that automated ticket workflows and cut decision time by 25%. He built ML-backed scoring models that improved financial forecasting by 5%, and architected entire data warehouses from scratch. At Zenith he built async API pipelines that doubled system performance.

Xi decides how Sato should think. Michael decides how Sato actually runs. He owns the infrastructure, the data pipelines, and the engineering that makes everything work in production. When Sato responds in three seconds instead of ten, or when Scam Shield handles a thousand checks without breaking, that is Michael's work.