A macro is a saved reply that support agents insert into conversations to avoid retyping the same answer. In Zendesk it is called a macro. In Intercom, a saved reply. In Front, a template. The name varies, the function is the same: pre-write the common answers, hit a shortcut, send.
Macros were the classic productivity tool of helpdesks. A well-maintained library of fifty to two hundred macros could double an agent's throughput on routine questions.
Why macros worked
The economics were simple. Most support questions repeat. Returns, shipping status, password reset, sizing questions, account access, refund timing. If twenty percent of your tickets are "where is my order", you write that reply once, save it as a macro, and every agent uses it. The variation between agents shrinks. The speed goes up. The customer gets a consistent answer.
Macros are still in use everywhere. They were honest tools that solved a real problem and continue to do so on teams without AI.
Why AI agents make most macros obsolete
A macro is a static reply. The customer says they want to know where their order is, the agent picks the macro, the macro says "your order ships within two business days", which is correct except today the customer ordered three days ago and the order is delayed because the warehouse had a power cut.
The macro does not know that. The agent has to read the situation, possibly edit the macro, possibly write something fresh. An AI agent retrieves the relevant context, including the order status and any delay notes, and writes the reply that fits this specific case. The macro is a template. The AI is a draft.
For straightforward repeating questions where context does not matter, macros are still fine. For most other things, AI replies are better.
Macros and AI together
The middle ground is macros as guardrails. The AI generates a reply, the team reviews and approves it for a category, and from that point on the AI uses that exact wording for that category. This way you get AI's contextual intelligence and the team's exact approved phrasing. The "macro" becomes the AI's voice template for a class of question.
This is how some teams transition off pure macros: keep the wording you trust, let the AI handle the variations.
In Keloa
In Keloa, the AI agent replies are generated fresh per conversation using RAG and grounding. Teams that want approved phrasing can lock in templated openings or closings. The rest of the reply adapts to the case. See how the AI works.