Articles for people who
ship the stack
Rewrites on React, Node.js, TypeScript and AI. A translated body is shown when it exists. If it does not, you see the English text and a note that says so.
Tagged: agent-memory
Questioning the Agent Stack: Vector Stores, Hypergraphs and Coding Jobs
Three cases where a bigger agent stack missed the real bottleneck: file-based agent memory beat a vector store, hypergraph RAG, and what senior work is.
12 min readFour Memory Strategies for AI Agents: Rolling Windows, FAISS and GraphRAG
Compare ways to give an LLM agent memory, from full history and rolling windows to inverted indexes, FAISS semantic search and a GraphRAG knowledge graph.
7 min readSix Agent Memory Tools Ranked by One Test: Can They Handle a Changed Fact?
Six agent memory tools face one test: can they handle a changed fact? Zep, Cognee, Mem0, Letta, LangChain Memory and plain vector stores, and where each fits.
7 min readClaude Code Context and Memory Internals, Rebuilt as a LangGraph Agent
How Claude Code builds API calls, runs subagents and compacts long sessions, then the same ideas rebuilt in LangGraph with checkpointers and memory files.
11 min readAI Agent Memory Is Not a Database: What to Keep, Retrieve and Forget
Why agent memory is a relevance problem: how context, memory, state and RAG differ, what to store, how to scope, age and audit it, and why forgetting is a feature.
11 min readAI Agent Memory in 2026: Mem0, Letta, Zep and the Cloud Platforms Compared
A tour of agent memory in 2026: the CoALA memory types, Mem0, Letta, Zep, cloud offerings, benchmarks, poisoning risks, and which design fits which job.
30 min readA Four-Layer Memory Design for AI Agents With Provenance
Short and long-term memory split, four layers (buffer, episodic, semantic, procedural), storage choices, async writes, and defenses against memory poisoning.
23 min readWhy Agent Memory Benchmarks Mislead, and How to Evaluate on Your Own Data
Agent memory benchmarks mislead with flawed answer keys and graders. Test retrieval apart from answers, add recency, abstention and forgetting checks.
15 min readWhy a Coding Agent's Memory Should Hold Only the Delta, and How to Store It
Coding agent memory should hold only what the model couldn't guess. Four tests for every entry, three needs for the store, and why markdown in git fits well.
18 min read
About these articles
Request a 24h estimate
Need the same stack in a production operator layer? Send the brief — estimate within 24 hours.