architecture

Multi-agent

Multi-agent refers to an architecture where several specialized AI agents collaborate — a router decides who takes over, and they hand off tasks cleanly.

In detail

Instead of one 'mega-agent' meant to do everything, we build small, focused agents. Benefits:

Higher per-agent quality (smaller context, clearer persona)
Easier to test (each agent in isolation)
Independently optimizable (support agent can run a different model than sales)
Clean escalation to humans

We usually run: router agent (detects intent) → specialized agents (support, sales, ops) → tools.

Example

User writes 'Considering Pro, but also where's my last invoice?' — router detects: sales agent first (Pro), then ops agent (invoice). Both respond coordinated in one message.

Related terms

Embedding
An embedding is a numeric representation of text (or an image) in a high-dimensional space where similar content sits close together — the foundation of semantic search and RAG.
RAG
RAG (Retrieval-Augmented Generation) is a technique where the language model retrieves relevant documents from your knowledge base before each answer and uses them as the basis for its response — so the agent stays current and answers with sources.
Tool use
Tool use (or function calling) is a language model's ability to call external functions with structured arguments — e.g. an API, a database query, or a skill.