Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore the critical role of memory in AI agents through this 28-minute technical podcast episode that addresses one of the most challenging aspects of building intelligent, conversational AI systems. Learn why traditional "digital goldfish" agents fail to maintain context across interactions and discover how to implement both short-term session-based memory and persistent long-term memory solutions. Master the fundamentals of agent memory architecture, starting with basic session management for single conversations before advancing to sophisticated long-term memory systems using Vertex AI Memory Bank, a managed service that enables agents to remember and learn from past interactions like humans do. Understand the key differences between Memory Bank's agentic approach to memory management and traditional RAG systems, including how LLMs intelligently decide what information to extract, update, or delete from memory stores. Dive deep into Memory Bank's core processes of memory extraction and consolidation that ensure stored information remains meaningful, deduplicated, and current. Get practical guidance on integrating Memory Bank with popular agent frameworks including ADK, LangGraph, and CrewAI through straightforward API implementations. Distinguish between sessions (conversation history) and memories (self-contained meaningful facts) while learning how to implement scoped memory systems for team collaboration and multi-user access. Stay current with the latest developments in the AI agent ecosystem, including new models like Zhipu AI's GLM-4.5 optimized for agentic tasks, LangChain's Align Evals evaluation framework, and updates to the Agent Development Kit and Agent-to-Agent protocol. Access comprehensive resources including official documentation, hands-on tutorials, and community forums to continue your agent development journey.
