Glossary
The layer of tooling and logic that manages how a large language model is called, chained with other tools and data sources, and integrated into a broader application workflow — handling prompt construction, retries, tool-calling, and multi-step reasoning chains.
Why It Matters
A raw call to a language model returns one response to one prompt — it doesn't know how to look up current data, call another system, or break a complex task into steps on its own. Orchestration is the engineering layer that turns a model into a system capable of actually completing a multi-step task reliably.
This is part of our AI Systems capability — the infrastructure that lets intelligence get embedded into a workflow, not bolted on as a standalone feature.
See the full AI Systems capabilityIs LLM orchestration the same as prompt engineering?
Prompt engineering is one piece of it — orchestration is the broader system around the model: tool-calling, retries, chaining multiple calls together, and managing state across a multi-step interaction.
Do all AI features need an orchestration layer?
Simple, single-turn use cases often don't need much. Orchestration becomes necessary once a workflow requires the model to use tools, retrieve external data, or complete a task across multiple steps.