[M2L 2025] 4.1 Frontiers of Agentic AI - Chi Wang

[M2L 2025] 4.1 Frontiers of Agentic AI - Chi Wang

🎙 Chi Wang 👥 3K 📅 November 13, 2025 ⏱ 71 min 👁 45 📄 expert opinion 🧭 2026-08-15
Available in: English (current) Français

Keywords

agentic AImulti-agentAutoGenAG2conversation programming

Summary

Chi Wang, creator of AutoGen/AG2, presents a talk on the frontiers of agentic AI at the Mediterranean Machine Learning summer school. He begins by outlining his vision of an ideal AI agent that can learn and collaborate, then introduces the core concepts of AG2: agents as primitive units with different backends (LLM, tools, human input) and conversation programming patterns (sequential, nested, group chat) to orchestrate them. He illustrates these concepts with a customer service example and discusses the evolution of the project from AutoGen to AG2. The talk highlights several production use cases, including chip design at NVIDIA, marketing at Walmart, and investment analysis at Better Future Labs, emphasizing the importance of domain expertise and hierarchical structures. Wang concludes by discussing remaining challenges and future directions for agentic AI research.

130 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides valuable insights into the practical design and deployment of multi-agent systems, drawing on real-world examples from NVIDIA, Walmart, and Better Future Labs. The argumentation is based on the speaker’s extensive experience and case studies, which adds credibility but lacks formal empirical evidence. The emphasis on conversation programming and hierarchical structures offers a useful framework for building complex agents. However, the talk is more of an expert opinion than a rigorous scientific presentation, with limited critical analysis of limitations or alternative approaches.

93 words

Title / Content Match

The title accurately reflects the content, which explores the frontiers of agentic AI through the lens of the AG2 framework.

Quality & Reliability

8/10

The speaker is a recognized expert in the field, having created AutoGen/AG2, and provides concrete examples and lessons from real-world deployments. However, the talk is largely anecdotal and lacks formal citations or peer-reviewed evidence, relying on personal experience and case studies.

Key Moments

Cited Sources

Concurring Sources

Contribution & Novelties

The talk provides a practitioner’s perspective on building and deploying multi-agent systems, highlighting the importance of conversation programming and hierarchical structures. It offers concrete examples from industry, such as NVIDIA’s chip design and Walmart’s marketing, demonstrating the practical value of these approaches. The speaker also shares lessons learned, such as the need for clear consensus criteria and the role of domain expertise.

Pour aller plus loin :

101 words

Radar Profile

The radar profile shows high scores in quantity and quality of information, reflecting the speaker's expertise and the rich content. The technical level is moderately high, suitable for an audience with some background. The overall reliability is good but not perfect, due to the lack of formal citations.

Reliability 7/10