Sakana AI just launched Sakana Marlin, its first commercial product. It’s an autonomous research assistant aimed at business users. You give it a research topic, and Marlin runs unattended for up to eight hours, generating both summary slides and a detailed report of up to a hundred pages. The goal is to replicate the kind of deep strategic work a CSO and small team might take weeks to finish, by having the AI form hypotheses, gather data, verify findings and map causal links in complex business environments. Under the hood, Marlin relies on Sakana AI’s long-horizon reasoning tech and multi-model coordination methods. The company built on past projects like AI Scientist (published in Nature), AB-MCTS (NeurIPS 2025 Spotlight) and ALE-Agent to stitch together research insights with real-world deployments. Those systems let Marlin handle hypothesis generation, information collection and structured analysis without manual intervention. Before the public launch, about 300 professionals from banks, corporate strategy teams, consulting firms and think tanks took part in a closed beta from April 2026. Participants used Marlin for tasks such as market research, risk assessment and competitive analysis. They praised its ability to dig deeper than typical chat-based tools and pointed out needs around output formats and stability for long jobs. Sakana AI used that feedback to refine report structures, improve formatting and bolster task reliability. Marlin is available immediately via self-serve signup, with a free pay-per-use tier plus Pro, Team and Enterprise subscriptions. The company plans to keep integrating new research advances—optimizing model coordination and expanding beyond chat interfaces—while recruiting talent to push its AI solutions further. For pricing and regional availability, check the product page.