The hard part of AI in high-stakes work isn't the model.It's earning enough trust that someone will act on the output.
Nine years in product. Five building AI document automation for institutional financial services, seed through Series B, first as product hire then as Director of Product.
The agent reasons.
Deterministic rules validate.
The human decides.
Site map
About me
Takeshi builds AI systems for institutional financial services, where a wrong answer has real consequences. He spent five years at an AI document automation startup, first as its product hire and later as Director of Product, taking it from a seed stage prototype through a Series B. Before that he led a small forward deployed engineering team at Booz Allen Hamilton, building a machine learning database analysis tool for a federal client. He writes the specs, reviews the code, and owns the product decisions himself.