TOKYO STARTUP BUILDS AI RESEARCHER THAT THINKS FOR 8 HOURS STRAIGHT AND WRITES 100-PAGE REPORTS Sakana AI, a Tokyo-based startup valued at $2.65 billion, launched Sakana Marlin this week, an autonomous research agent that the company is positioning as a virtual chief strategy officer. The pitch is unusual by the standards of the current AI product landscape. Most AI research tools produce quick summaries. Marlin produces reports that can run over 100 pages and are built over eight continuous hours of autonomous work. The system forms its own hypotheses, gathers information from online sources, identifies conflicting evidence, reconciles it, and synthesizes the result into structured deliverables including executive slides. It is not a chatbot that runs a search and summarizes the results. It is closer to what a senior consultant might produce after weeks of focused analysis, compressed into a single overnight run. Sakana ran a closed beta with approximately 300 professionals from finance and consulting before launching. The pricing reflects the target market. Users pay either per run or through monthly subscriptions starting at roughly $1,000. The timing matters. Enterprise AI spending is accelerating and companies are actively looking for AI products that can handle genuinely complex knowledge work rather than just generating drafts and summaries. Sakana Marlin is a direct bet that the market is ready to pay for AI that replaces senior-level intellectual labor, not just clerical work. Keywords: Sakana Marlin AI agent, Sakana AI research tool, autonomous AI researcher, enterprise AI deep research