Agents

Agent memory is the state an AI agent keeps across steps or sessions, such as scratchpad notes, retrieved facts, user preferences, or task history.

Development

Context engineering is the practice of designing what information, tools, memory, and instructions are supplied to a model at runtime.

Inference

RAG retrieves relevant passages (often via embeddings) and conditions generation on them—reducing reliance on parametric memory alone.

Training

Retrieval-augmented generation (RAG) grounds answers on retrieved documents instead of parametric memory alone.