Book: AI That Plans The series: AI in TypeScript — 5 books, from your first LLM call to agents in...
Three separate memory layers—working, long-term, episodic—each with distinct lifetimes and access patterns, replace the common single-array approach causing arbitrary truncation. Separating stores prevents stale facts from crowding recent turns, improving agent stability in production.
Book: AI That Plans
The series: AI in TypeScript — 5 books, from your first LLM call to agents in production — all five here
My project: Hermes IDE | GitHub — an IDE for developers who ship with Claude Code and other AI coding tools
Me: xgabriel.com | GitHub
"Agent memory" gets implemented as one array of messages that grows
Three concrete memory patterns with a schema for each, and the reason the third one — plain files — outperforms the vector store…
If you have ever built an autonomous AI agent, you have likely run into the "memory bloat" problem....
Agent memory explained: the 7 types — working, semantic, episodic, procedural, retrieval, parametric, prospective — and when to…
Understanding different memory architectures for AI agents and when persistent, episodic, or vector memory actually pays off in…
Turn 30 and the window is at capacity. Which parts of an agent's history are safe to drop, which must survive, and how to compact…
Most teams building AI agents are treating memory and inference cost as something the next model...
