Episode 4- Memory in AI agents (short-term vs long-term)

Most AI systems don’t actually “remember” anything. Unless memory is deliberately engineered, every interaction starts from zero.
In this episode, we break down memory in AI agents, specifically the difference between short-term memory inside a model’s context window and long-term memory stored externally in databases or vector stores. You’ll understand what each type does and why both are necessary for real-world agents.
We explain how memory is stored, retrieved, and reinserted into the model’s reasoning process, and where systems typically fail.
By the end, you’ll be able to evaluate agent architectures, design for persistence, and make informed product decisions around personalization, continuity, and cost.