AI Operations · Glossary
Macro
Component 5 of the Eight Components of an Interactive Agent. A Macro is a reusable block of authored message text, written once and interpolated wherever the agent needs consistent wording, governed centrally and owned by the domain expert who authored it.
A Macro is the agent's reusable phrasing, not a behaviour. Where a Policy governs a decision and a Routine governs a flow, a Macro standardises the words: a canned disclosure, a fixed escalation message, a regulator-approved wording the operator must reproduce verbatim. The domain expert authors the block once and references it by name from Prompts and Routine steps, so the same text renders identically everywhere it appears. Macros are not compositions of Routines and Policies, and they are not infrastructure. They are content assets, versioned inside the policy library, governed under one accountable function. Authoring wording once prevents the drift that comes from re-typing approved language across dozens of touchpoints, and every rendered Macro is inspectable per decision in the production trace.