03 / Working notes

Blog

Notes on using, building, and testing AI systems.

AI Memory Layer: Designing Persistent Memory for Agents

What should an AI agent remember? A practical look at scoped memory, human sources, retrieval, disagreements and where a graph helps.

When your agent's state gets complex, full replacements break. Learn how to use incremental operations with CQRS reducers in LangGraph.

Evaluate an AI proposal before hiring: compare alternatives, test real tasks, review data access and costs, and define when to approve a pilot.

An introductory explanation of the concepts to consider when implementing short-term and long-term memory in agents.

How to decide what an AI agent should enable in your product: user needs, knowledge, actions, evaluation, and the boundaries of a first version.