Cutting-edge AI Technology for Business: Sakana AI Applied Team Member Interview

Sakana AI built its reputation on academic firepower — papers like AI Scientist and Darwin Gödel Machine, work that leans on ideas borrowed from nature. In 2025 the company launched an Applied Team to push that research into actual business use, starting with finance and the defense and public sectors. The team has grown to roughly 35 people in a short stretch, pulling in engineers and business specialists from major domestic tech firms.

Two members explained how the work actually happens. Kano, an Applied Research Engineer who joined in April, spent his first weeks building a prototype for a finance project — and he's already one of the more senior engineers on the team, simply because it's that new. Before Sakana, he built consumer-facing systems for a taxi-dispatch app and a live-streaming service, while finishing a PhD in AI research on the side. He describes wanting to be someone strong in both research and real-world deployment, and says Sakana's shift toward commercialization fit that ambition.

Kano is particularly drawn to what he calls open-ended exploration — the kind of continuous novelty-seeking baked into projects like AI Scientist. He thinks that as automation handles more routine work, the next frontier is AI that helps people generate ideas and map out future possibilities: strategy discussions, threat assessments, that sort of thing. What makes the setup work, he says, is that Sakana hires actual domain experts into finance and public-sector roles, so engineers who start with little knowledge of those fields get daily feedback loops from people who know the terrain.

Project manager Nakagawa, who came from an economics background and did data-analysis internship work before joining a bank during the fintech boom, offers the client-facing view. He recalls biometric authentication accuracy improving roughly tenfold in a single year back then, which hooked him on watching technology jump generations. An MBA stint led him to the NeurIPS conference in 2022, right around ChatGPT's release, and an internship at Sakana sealed the deal — he points to founders David Ha, Llion Jones and Ren Ito as part of what convinced him this was a once-only moment. When he joined, the business side had just two people; it's now around 15.

His current project automates part of the work behind client proposals in finance, an attempt to apply the same AI Scientist logic — a machine producing a finished, consensus-worthy output — to non-technical, so-called 'humanities' work instead of research papers. He offers a concrete illustration: a proposal that once took weeks could, in theory, be turned around in two hours, fast enough to respond to something like a tariff announcement almost immediately. Nakagawa also notes that about 80 percent of the business-side staff can write code themselves, opening Visual Studio Code alongside engineers to fix things directly — a habit he sees as central to how fast the team iterates.