Skip to main content
We use a five-section prompt structure across our production agents because it is the most reliable way to build agents that perform consistently on live calls. Each section has a clear purpose. Role defines the agent’s identity, tone, and objective. It answers three questions: who the agent is, how it should sound, and what it is trying to accomplish on every call. Context contains the business details, caller information, and dynamic variables the agent needs before the conversation begins. This includes things like the contact’s name, when they enquired, what they were interested in, and any other relevant data passed into the call. Availability defines business hours, booking rules, and any scheduling restrictions. This section prevents the agent from offering times that should never be available, such as weekends, holidays, or hours outside your operating window. Guidelines control how the agent behaves throughout the conversation. This is where you define conversation style, objection handling, response length, tone rules, and how the agent should deal with edge cases. It is the section that turns personality into repeatable behaviour. Main Script is the ordered flow of the conversation from introduction to close. It defines what happens first, what happens next, and how the agent should move through the call based on the contact’s responses. In most production agents, this is the single biggest driver of consistency and conversion rate. You can also add Handlers as an optional layer. Handlers are named response blocks for specific situations, such as when the contact is busy, asks whether the caller is AI, becomes hostile, or when a booking fails. They keep the main script cleaner and make edge-case behaviour more predictable.