As generative AI evolves beyond static prompts, this summit brings together top researchers, builders, and strategists to explore the future of goal-directed agents built on open source and open standards. The Open Agent Summit on October 21, co-located with PyTorch Conference in San Francisco, is the premier event focused on the rise of agentic AI, intelligent, autonomous systems capable of perceiving, planning, and acting in dynamic environments.
Core themes include multi-agent collaboration, tooling and infrastructure, protocols and standards, real-world deployment, safety and alignment, and the integration of agents across enterprise and consumer applications.
Top 3 Reasons to Attend
- Get in on the cutting edge of agentic AI – This summit is specifically focused on goal-directed agents – AI systems that perceive, plan, and act in changing environments, not just static prompt-responses. If you’re interested in where AI is going beyond large language models into more autonomous, interactive, and adaptive systems, this event will be full of the latest research, frameworks, and real-world deployment stories.
- Open source, open standards & multi-agent collaboration – One of the core themes is how to build agents using open source tools and open protocols, as well as standards that make agents interoperable. You’ll hear from experts working on infrastructure, tooling, multi-agent systems, and standards. For developers, researchers, or strategists, this is a chance to contribute to, and learn from, the community working to establish the foundations of responsible, scalable, and collaborative agentic systems.
- Real-world deployment, safety & alignment – Beyond theory, the summit will cover practical deployment issues: safety, alignment, reliability, infrastructure, how agentic AI can or should work in real-world settings (enterprise and consumer), what protocols are needed to ensure trust, how tooling & infrastructure can support safe agents, etc. Whether you build agents, use them, or are thinking about their impact, attending this summit gives you direct access to insights, case studies, and thought leadership from the industry leaders building and deploying agents in production.
Program Highlights
Through keynotes, panels and deep technical sessions, the summit offers a comprehensive look at how open autonomous agents are reshaping industries, workflows, and human-machine interaction.
Keynotes
- Welcome & Opening Remarks – Matt White, Executive Director, PyTorch Foundation
- Towards Building Safe and Secure Agentic AI – Dawn Song, Professor, UC Berkeley
- Building an Open Agentic Future – Samuel Colvin, Founder, Pydantic
Sessions
- AG2/MassGen – Chi Wang, DeepMind
- AI Scientist – Yutaro Yamada, Sakana AI
- Building and Sharing Reliable AI Agents with BeeAI – Sandi Besen, IBM
- A2A – Chitra Venkatesh, Google
- Using LlamaIndex to Build an Open-Source NotebookLM Alternative – Tuana Çelik, Llamaindex
- The Model Context Protocol: Connecting AI to Everything – Adam Jones, Anthropic
- Accelerating Open-Source RL and Agentic Inference with vLLM – Michael Goin, Red Hat
- AGNTCY: The Open Source Project for Building Multi-Agent Systems – John Parello, Cisco
- A Modular RL Library for LLM Agents – Sumanth Hegde, Anyscale & Tyler Griggs, UC Berkeley
- End-to-End Optimizing Multi-Turn RL and High-Performance Inference in Agents with SGLang – Chenyang Zhao, SGLang
Panels
- The Future of Agents and Human-Agent Collaboration
- Yujian Tang, OSS4AI (Moderator)
- Davide Crapis, Ethereum Foundation
- Mitesh Patel, NVIDIA
- Bradley Axen, Block
- Yu (Jo) Zhou, Columbia University
- Agents in the Wild: Real-World Deployments and Lessons Learned
- Yujian Tang, OSS4AI (Moderator)
- Yina Arenas, Microsoft
- Andrew Hoh, Lastmile AI
- Diamond Bishop, DataDog
- Mahdi Ghodsi, AMD
Register by adding the Open Agent Summit to your PyTorch Conference registration.