Sakana AI and Daiwa Securities Move Joint AI Project to Production Development Phase for Wealth Management

Tokyo-based Sakana AI just cleared a milestone that a lot of enterprise AI pilots never reach: it's moving from proof-of-concept into actual production. The company signed a partnership with Daiwa Securities back in September, and the two spent the months since testing whether Sakana's agent tech could handle something as unglamorous but essential as gathering and analyzing market information for financial consultants.

The pilot leaned on Sakana's homegrown tools, including its AI Scientist system and a search technique called AB-MCTS, to build an agent pipeline that could collect data, digest it, and hand consultants something usable. According to the companies, it worked across every stage of the workflow, not just the easy parts, and it held up on both quality and processing speed. They also found a way to fold in feedback from actual users so the system keeps improving rather than staying frozen at launch quality.

That's enough for Daiwa to greenlight a production phase starting August 1, 2026, this time aimed squarely at wealth management. The goal is blunt: give financial advisors more time with clients and less time buried in research, while making the advice itself sharper. Japanese retail investors are increasingly diverse in what they want from asset management, and Daiwa is betting that AI-assisted research is the way to keep personalized advice from becoming a bottleneck.

What's notable here isn't the technology itself, agent systems doing research synthesis are common by now, but the fact that a major securities firm is willing to stake a customer-facing consulting product on one. Sakana AI, which has positioned itself as Japan's answer to the big Western AI labs, gets a real-world validation story that goes beyond benchmarks. If the rollout inside Daiwa goes smoothly, expect this partnership to become the template Sakana points to when it goes after other financial institutions.