Tokyo-based startup Sakana AI has launched its first commercial product, an autonomous Sakana Marlin tool designed for business research.
The tool operates independently for up to eight hours to generate structured summary slides and detailed reports. Consequently, it aims to automate complex strategic analysis tasks that typically require weeks of human effort.
Users only need to set the initial research topic. After that, the artificial intelligence system autonomously formulates hypotheses, gathers information, and refines its findings. Notably, this process allows the tool to produce reports spanning up to one hundred pages.
Capabilities of the Sakana Marlin tool
The Sakana Marlin tool targets corporate strategy teams, financial institutions, consulting firms, and think tanks. Meanwhile, the system executes thousands of large language model queries per session. This high volume of queries helps the tool analyze deep strategic questions rather than delivering quick, surface-level summaries.
Technical Architecture and Development
The technology relies on Adaptive Branching Monte Carlo Tree Search, also known as AB-MCTS. Specifically, this method coordinates multiple models to solve complex problems through trial and error. Furthermore, the workflow automation framework originates from the Sakana AI Scientist project, which was published in Nature.
The development team balanced two search strategies during the design phase. The system can search deeper by refining a promising answer, or search wider by generating new solutions. As a result, the tool dynamically directs computational resources toward the most promising research paths.
Beta Testing and Market Strategy
A closed beta phase began in April 2026 with approximately 300 professionals from various industries. These participants represented financial institutions, operating companies, and consulting firms. According to the company, testers reported that the Sakana Marlin tool was highly practical for deep information gathering.
The product is positioned as infrastructure for modern decision-making in volatile markets. Traditional strategic research often takes weeks, which can make insights outdated. Therefore, the company designed this tool to bridge the gap for teams operating in fast-moving environments.
Financial Background and Future Plans
Sakana AI was founded in Tokyo in July 2023 by David Ha, Llion Jones, and Ren Ito. Recently, the company raised 135 million USD in a Series B funding round, reaching a valuation of 2.65 billion USD. This funding round closed on November 17, 2025, with participation from major global investors.
Investors in the startup include Mitsubishi UFJ Financial Group, Khosla Ventures, and Salesforce Ventures. Additionally, Citigroup made an investment, marking its first such transaction in Japan. The company plans to expand its economy and enterprise business into the manufacturing and government sectors in late 2026.
Source: X (@SakanaAILabs) — researched
