Welcome to “AI in the Real World"! I’m your host, Joanne Chen, a General Partner at Foundation Capital, where I work closely with startups that are reshaping business with AI. In this series, I’ll be holding in-depth discussions with leading AI researchers. Together, we’ll explore how state-of-the-art AI models are being applied in the real world.

To kick things off, I'm excited to speak with Joon Sung Park, a Ph.D. student in computer science at Stanford. Joon works at the intersection of human-computer interaction, natural language processing, and machine learning. He's best known for his pioneering research on generative AI agents.

We break down how generative AI is transforming agent design, share advice for builders working with these models, and unpack why we haven’t yet found the "killer app" for LLMs.

Read the top takeaways from our conversation here: https://foundationcapital.com/the-fut...

(00:00:00) Introduction
(00:04:00) The impact of LLMs on agent capabilities
(00:08:07) Tool use and simulations are the two main focuses of agent R&D
(00:10:34) Multi-agent systems and their applications
(00:14:15) LLMs represent a new paradigm for agent design
(00:18:55) Multimodal models promise step-change improvements in agent performance
(00:23:55) Unsolved challenges: grounding, representativeness, and scalability
(00:26:40) Soft-edged problems are better bets for AI builders
(00:33:00) The future of transformer architectures
(00:38:49) Why agents need to solve real user needs
(00:39:59) We haven’t yet found the “killer app” for LLMs
(00:44:47) Classic product principles still apply when building with LLMs