AutoGen (AG2)
Web3 / ai data
AutoGen, originally developed by Microsoft and now maintained as AG2 by the open-source community, is one of the most widely deployed multi-agent AI frameworks, built on a fundamentally conversational orchestration philosophy. Rather than modelling workflows as graphs or task queues, AutoGen treats multi-agent collaboration as a structured message exchange: developers define agents with specific roles (e.g., a 'Coder' agent and a 'Reviewer' agent), and those agents exchange messages in a dynamic loop to delegate tasks, debate solutions, critique outputs, and reach consensus. This conversational model naturally accommodates code execution and tool use within the dialogue, making complex reasoning chains human-readable and auditable. AutoGen has seen massive enterprise deployment across data science, research, and software engineering workflows and is the standard framework for teams that prefer to coordinate AI collaboration through structured dialogue rather than rigid routing. Example: A blockchain analytics team builds an AutoGen pipeline where a 'Researcher' agent queries Dune Analytics for on-chain data, a 'Analyst' agent interprets the patterns and drafts conclusions, and a 'Critic' agent stress-tests the reasoning and flags weak inferences — the entire research cycle running as a structured multi-agent conversation that produces a peer-reviewed report. Why it matters for AI and data in Web3: On-chain research and protocol analysis benefit from multi-perspective reasoning. AutoGen's conversational architecture makes it straightforward to assign different LLMs with different strengths to different agent roles — using a reasoning-optimised model for analysis and a safety-focused model for compliance checks — within a single auditable workflow.
Explore the full Web3 Glossary — 2,062+ expert-curated definitions. Need guidance? Talk to our consultants.