Session brief
Stateful Agents — Full Workshop with Charles Packer of Letta and MemGPT
Overview
This workshop introduces the concept of stateful agents in AI, emphasizing that true agentic capabilities require memory and the ability to learn from experience, which current stateless Large Language Models (LLMs) lack. The session explores how to build these stateful agents using the Leta framework, focusing on memory management systems and tool-calling mechanisms to create more human-like and persistent AI interactions.
Who should watch
- AI engineers and developers building agentic applications.
- Product Managers exploring the potential of advanced AI agents.
- Builders seeking to create AI assistants, companions, or co-pilots.
- Those interested in overcoming the limitations of stateless LLMs in agent development.
- Individuals looking to understand and implement memory management for AI agents.
Key takeaways
- Current LLM-based agents are fundamentally stateless, requiring explicit mechanisms for state management and memory to achieve human-like intelligence and learning.
- Stateful agents, or agents with memory, are crucial for building sophisticated AI applications that can learn, adapt, and provide personalized experiences over time.
- The Leta framework provides a server-client architecture designed for stateful agents, enabling persistent memory and session-based interactions without needing to resend full conversation history.
- Memory in Leta agents is managed through memory blocks, which can be core memory (within the context window) or archival/recall memory (external data sources), allowing agents to store, retrieve, and modify information.
- Tool calling is central to Leta agents, enabling them to interact with external systems and manage their own memory, with mechanisms for tool execution on the server-side.
- The framework supports advanced features like context window management to control token usage, custom tool creation, and multi-agent communication through asynchronous and synchronous message passing.
- Building stateful agents involves careful prompt engineering and configuring memory structures to influence agent behavior and ensure effective learning and adaptation.
Notable quotes
*The fundamental unit of compute we're using for AI is stateless. It's not a recurrent neural network. It's a transformer. The transformer inherently is like a stateless machine.*
*Stapleness or memory is actually probably the most important thing to solve if we actually want to get you know if you want 125 be the year of agents.*
*The promise of stateful agents: the experience should get better and better and better over time as the AI kind of learns more and more about you.*
Unofficial community note. Prefer the recording for nuance.