Sakana’s Fugu is a multi‑agent orchestration platform that claims to outperform leading large models by dynamically routing tasks among specialized agents, but its marketing narrative, benchmark claims, case studies, cost, latency, and transparency raise significant technical and governance questions.
The article details how an AI‑augmented workflow completed a 46‑page research paper in six days using 108 agent calls and 648 k tokens, introduces an L1‑L5 autonomy taxonomy, compares four architectural patterns across 17 systems, and highlights six open challenges and key bottlenecks such as continual knowledge accumulation and reliable self‑assessment.
The article summarizes two recent DeepSeek papers—one presenting an L1–L5 taxonomy and four architecture patterns for autonomous research agents, the other proposing a three‑dimensional taxonomy for continual learning, detailing method families, a self‑improvement phase diagram, experimental comparisons, an impossibility theorem, and the production statistics of the Deli AutoResearch framework.
