Getting Up to Speed on Multi-Agent Systems

An eight-part series mapping the multi-agent LLM literature for readers coming in fresh. Two waves of research, one outside disruption that reshaped both, and the questions the field is still trying to answer.

Who this is for: you already ship or evaluate LLM agents (tools, long context, basic eval loops) and want the research landscape in view. This is not an on-ramp to transformers or prompting fundamentals.

Read straight through, or jump in wherever the title pulls you. The first two parts set up vocabulary; parts 3–6 walk through specific clusters of papers; parts 7–8 zoom out to benchmarks and what's still open.

  1. Part 1. The Landscape April 24, 2026
  2. Part 2. The Vocabulary April 25, 2026
  3. Part 3. Wave 1: Can Agents Coordinate At All? April 26, 2026
  4. Part 4. Wave 2: Why It Breaks publishes April 27, 2026
  5. Part 5. Debate, State, and Coordination publishes April 28, 2026
  6. Part 6. Verification Patterns publishes April 29, 2026
  7. Part 7. Benchmarks and What They Miss publishes April 30, 2026
  8. Part 8. Open Questions publishes May 1, 2026

← Home · Full archive →