Specialist agent
A specialist agent in a multi-agent system is scoped to a narrow domain — code review, data analysis, document drafting, web research — with a system prompt, tool set, and context window tuned to that domain. The orchestrator delegates well-defined sub-tasks to specialists rather than handling everything itself.
Specialisation produces measurable wins because narrower scope allows tighter optimisation. A specialist's system prompt can be longer and more detailed than a general-purpose agent's; its tool set can include domain-specific tools (a coding specialist has run_tests; a research specialist doesn't); its few-shot examples can be sharper. The trade-off vs general agents is interface overhead — the orchestrator must know which specialist to call and how to phrase the request. The pragmatic threshold for spinning out a specialist: when the same kind of sub-task appears repeatedly and a tailored prompt would measurably outperform a generic one.
Related terms
- Multi-agent system
A multi-agent system uses multiple LLM-driven agents working together — typically with role specialisation (planner, executor, critic) or domain specialisation (researcher, coder, writer).
- Orchestrator agent
An orchestrator agent is the top-level controller in a multi-agent system — it decomposes the user's request into sub-tasks, dispatches them to specialist agents, integrates their outputs, and handles cross-cutting concerns like error recovery and progress reporting.
- Agentic workflow
An agentic workflow is a software process in which one or more LLM-driven agents drive the work, making decisions and invoking tools at each step rather than following a hardcoded script.