Thought Leadership Dinner "The Role of Memory and Retrieval in Agentic AI"
- Date
- 2025-06-19
- Location
- San Francisco, CA, USA
- Host
- Unusual Ventures
About this event
Memory is quickly becoming one of the defining questions in agentic AI. If agents are expected to plan, adapt, collaborate, and operate with more autonomy, then how they remember, retrieve, and use context is no longer a side topic, it is central to product design, model behavior, and real-world reliability. This dinner is built for people who want a sharper conversation than the usual high-level AI networking event. In a small-group, in-person setting in San Francisco, we will dig into how memory and retrieval shape agent performance, where current approaches break down, and what practitioners are learning as they move from demos to systems people can actually trust. About the Event Thought Leadership Dinner "The Role of Memory and Retrieval in Agentic AI" is an in-person evening gathering for people thinking seriously about the next layer of AI systems. The focus is not just on models in isolation, but on the infrastructure and design choices that allow agents to persist context, recall the right information at the right time, and act in ways that feel coherent over longer tasks. The dinner format is meant to support a more substantive exchange than a panel or lecture. Sharing a meal creates room for discussion that is candid, practical, and nuanced. You can expect the conversation to move beyond broad predictions and into the decisions teams are making right now around retrieval strategies, memory architectures, evaluation, user experience, and operational tradeoffs. This event is also about community. Agentic systems sit at the intersection of research, product, engineering, and design, and the most useful conversations often happen when those perspectives are in the same room. If you care about how autonomy actually works in practice, this is a setting designed to help you compare notes with peers who are asking similarly hard questions. What to Expect The evening begins at 6:30 PM PDT on Wednesday, June 18 in San Francisco. The flow is intentionally simple: arrive, meet the room, settle in over dinner, and move into a guided conversation around the technical and strategic role of memory and retrieval in agentic AI. You should expect a discussion-centered format rather than a one-way presentation. Topics may include questions such as: What kinds of memory matter most for agents: short-term context, long-term user knowledge, task state, tool history, or environment memory How retrieval affects planning, tool use, personalization, and multi-step execution Where current retrieval pipelines fall short in latency, relevance, grounding, or cost How teams are thinking about evaluation when an agent's behavior depends on what it remembers What product and trust implications emerge when agents store, forget, summarize, or reinterpret past interactions Because this is a dinner conversation, there is room for both big-picture thinking and practical examples. Some attendees may come from model, platform, or infrastructure backgrounds; others may be building user-facing applications. That mix is a strength. The goal is to surface useful patterns, disagreements worth exploring, and concrete insights that attendees can take back into their own work. Expect thoughtful introductions, focused discussion, and time to continue conversations more informally throughout the meal. If you have been looking for a forum where people can talk honestly about what is working, what is brittle, and what still feels unsolved in agent design, this event is built for exactly that. Why Attend The most interesting work in AI right now is increasingly shifting from single-prompt interactions to systems that need continuity. An agent that cannot reliably remember, retrieve, and reason over prior context will struggle to feel useful in anything beyond narrow tasks. This dinner is a chance to engage directly with one of the key design problems behind more capable autonomous systems. You will leave with a clearer sense of how others are framing the memory problem, what tradeoffs they are making, and how retrieval design affects downstream behavior. That could be valuable whether you are building agents, supporting teams that build them, or trying to understand where the field is moving next. There is also real value in the room itself. Events that focus on a specific technical and product question tend to attract people with sharper perspectives and better questions. Instead of general AI chatter, you can expect conversations with people who are actively wrestling with architecture, evaluation, product constraints, and user expectations. If you want signal over noise, peer learning over posturing, and a setting where dinner supports serious discussion rather than distracting from it, this event offers that balance. Practical Details This is an in-person event in San Francisco, USA. It is designed as a dinner gathering, so expect an evening format that combines conversation and food in a more intimate setting than a standard meetup. The event takes place on Wednesday, June 18 at 6:30 PM PDT. Since it is an in-person dinner, it is a good fit for local attendees or anyone already planning to be in San Francisco that evening. The tone of the event is thoughtful, grounded, and collaborative. You do not need to prepare formal remarks, but you will get more out of the evening if you arrive ready to contribute a perspective, a question, or a real challenge you are seeing in agentic systems. If memory, retrieval, and autonomy are active parts of your work or your curiosity, this dinner will give you a focused room, a timely topic, and the kind of conversation that is hard to get in larger, less specific AI events.
Who should attend
This is for people who want a sharper, more technical conversation about agentic AI than a typical networking event offers. - You are building or designing **AI agents, copilots, or autonomous workflows** and want to think more deeply about how memory and retrieval affect reliability, continuity, and user trust. - You work in **ML, infrastructure, data, or platform engineering** and care about the systems layer behind agent behavior, including context management, retrieval quality, latency, and evaluation. - You are a **product leader, founder, or researcher** exploring where agentic AI is heading and want practical insight into the tradeoffs behind more persistent, adaptive systems. - You are interested in **how agents handle long-running tasks, personalization, tool use, and multi-step reasoning**, and you want to compare approaches with others working on similar problems. - You value **small-group, in-person discussion** where people can speak candidly, ask better questions, and move beyond surface-level AI takes. - You are part of the broader **San Francisco AI community** and want to meet thoughtful peers around a specific topic that matters now, not just AI in general. If you have been looking for a room where the conversation starts at a higher level and quickly gets concrete, you will likely feel at home here.