Stanimira Vlaeva

AI agents can reason, call tools and perform complex tasks. However, when you start a new conversation, they behave as if they’ve never met you. Give them too much history and they get distracted, expensive and inconsistent. Give them too little and they forget user preferences or struggle with longer tasks. What is the right way to give the agent memory?

You can add a vector database, but that alone is not enough to solve the problem. You also need to decide what the agent should remember, when it should save something and how it should find that information later. Oh, and don’t forget that old memories can become stale - you have to figure out how to change or remove them and when.

Overall, building memory systems for AI agents is a data problem. This talk presents a practical way to think about this problem. We’ll look at working, short-term, semantic, episodic and procedural memory. Then, we’ll follow a memory through the system from creation to retrieval and eventual deletion. A short demo will compare exact, semantic and hybrid retrieval and show how the agent behaves with and without the right context.

The goal of this session is simple - help you build agents that remember useful information without filling their context with irrelevant or outdated details.

Stanimira Vlaeva is a Staff Developer Advocate at MongoDB and a Google Developer Expert for Google Cloud. She is passionate about explaining complex technical topics in an understandable way, live-coding and helping developers build with confidence.