The Memory Layer: Building the Next Generation of AI Agents
This intimate roundtable brings together a small group of founders, builders, and ecosystem leaders working across AI agents, memory, and developer infrastructure.
As AI systems become increasingly agentic, memory is emerging as a critical layer of the stack — shaping how agents retain context, learn from past interactions, and operate more effectively over time.
Over lunch, we’ll explore where the memory layer is heading, what builders are seeing in practice, and how it may shape the next generation of AI products and infrastructure.
We’ll also hear a few short opening perspectives from builders working directly on memory and agent infrastructure before opening up the roundtable discussion.
Opening Discussion
Dr. Zhiyu Li
Co-founder & CTO, MemOS
Zhiyu is the co-founder and CTO of MemOS, a memory operating system for large language models. He holds a Ph.D. in artificial intelligence and previously led algorithm R&D teams at Alibaba and Xiaohongshu. His research focuses on memory systems for LLMs, with more than 70 papers published at venues including NeurIPS, ICLR, and ACL.
Harsh Nigam
Co-founder & CEO, Fay
Harsh is the co-founder and CEO of Fay, an AI startup building consumer agents designed to help people get things done. A repeat founder and product builder, he has worked across AI, enterprise software, and consumer products, with a focus on turning advances in AI into useful, intuitive products.
The conversation will be intentionally small and discussion-driven, with an emphasis on candid exchange, shared perspectives, and meaningful connections.
Hosted by MemOS and friends. Space is limited to keep the conversation focused and interactive.
Tentative Agenda
Welcome & introductions
Lunch and opening perspectives
Roundtable conversation on memory, agents, and the evolving AI infrastructure stack
Open discussion and connections
The format will remain intentionally flexible to follow the most interesting questions in the room.