Scripting, Prompting & Information
Understand the differences between scripting, prompting, and informational inputs when configuring AVA scenarios. This guide breaks down each interaction type and highlights how to use custom values for personalization.
Overview
This document provides a complete breakdown of:
- The difference between scripting, prompting, and information fields
- A reference list of all customizable input fields used across AVA scenarios and roles
- Guidance on using custom values to personalize dialogue
Details
When configuring AVA outbound and inbound voice flows, there are three primary types of input fields you’ll use:
- Scripting – Fixed language that the AI speaks verbatim.
- Prompting – Flexible guidance that lets the AI dynamically generate a response.
- Information – Parameters or facts that shape how the AI behaves, but are not spoken directly unless referenced.
You can also use custom values to personalize what the AI says based on each contact’s data. These tokens are automatically replaced with real information during live calls.
1. Scripting: Structured, Controlled Dialogue
Scripting refers to explicitly written, predetermined phrases that AVA will say. Think of it as the “scripted lines” given to a human agent — designed to be consistent, repeatable, and compliant.
Scripts are typically used when:
- You need the AI to deliver precise messaging (e.g., legal disclosures, appointment confirmations, survey intros).
- You want to control the tone or style of a greeting, offer, or transition.
- The call needs to stay on-brand or within policy boundaries.
In AVA, scripted lines are often:
- Used at the start or end of a call (e.g., “Hi, this is Ava calling on behalf of…”)
- Included between actions like transfers or fallback steps
- Paired with response rules that restrict the AI’s freedom
Example:
“Just confirming, this is regarding your upcoming service appointment on Tuesday at 3 PM, correct?”
Scripting provides maximum predictability, but limited adaptability.
2. Prompting: AI-Driven Engagement
Prompting, by contrast, is about giving AVA guidance, not instructions. You describe what AVA should try to achieve — not how to say it word-for-word. The AI then generates responses dynamically based on the user’s input and the goal of the prompt.
Prompts are typically used when:
- You want the AI to engage naturally in a conversation.
- The flow depends on how the other person responds (e.g., answering questions, handling objections).
- You need flexibility in tone, length, or phrasing.
In AVA, prompts are written to set context and intent:
- “Ask the client to confirm their availability for a scheduled appointment.”
- “Politely explain the service and ask if they’re interested.”
Example Prompt:
“Politely explain that this is a follow-up about a previous inquiry and ask if they would like to proceed.”
The AI will then generate a voice response like:
“Hi! I’m just following up on your earlier request — would you like to move forward with that today?”
Prompting provides adaptability and personalization, but less predictability.
3. Information: Context That Powers Logic
Information fields are not spoken aloud by the AI, but they influence how it behaves behind the scenes. These values provide factual or contextual data that shape call logic, guide tone, or drive how dynamic content is inserted.
Information fields are typically:
- Parameters such as meeting length, profession, or location
- Used in custom values to personalize scripts and prompts
- Responsible for feeding conversation routing, calendar logic, and transfer workflows
They act as a foundation for call personalization — enabling AVA to sound informed without hardcoding every variation.
Example:
Setting the field “Profession = Real Estate Agent” may cause the AI to adopt appropriate language when referencing listings, showings, or clients.
While not spoken directly unless referenced inside a script or prompt, these fields are essential to maintaining contextual accuracy in every interaction.
What Are Custom Values?
Custom values are dynamic placeholders used to personalize AI speech by inserting real contact-specific information at runtime.
These values are embedded within user-defined fields like voicemail messages, introductions, or prompts.
| Custom Value | Description |
|---|---|
{{client_firstname}} | The contact’s first name |
{{appointment_time}} | Time of the scheduled appointment |
{{appointment_date}} | Date of the scheduled appointment |
{{representative_name}} | Human rep the AI is representing |
{{company_name}} | Name of your agency or company |
These tokens are automatically replaced during live calls. They’re not fields users fill out directly, but are available for insertion into other inputs.
Customizable User Input Fields
When configuring AVA scenarios, users define various fields that shape how the AI behaves, speaks, and responds — including introductions, calendar logic, objection handling, voicemail, and more. While many fields support scripting, prompting, or information input interchangeably, it’s important to focus on few uses for each field. The more clearly each field is structured and aligned with its purpose, the more effectively the AI will deliver consistent, smooth, and confident interactions.
Here are some suggested uses for each available field:
Inbound Calls
Outbound Calls
General Advice
1. Keep Scripting Concise and Purposeful
Avoid stuffing scripting fields with excessive information, long introductions, or hardcoded names. The AI should quickly address the point of the call — such as confirming a meeting or prompting action — in a natural, professional way. Excessive exposition confuses contacts and leads to longer calls.
Remember what you’re buying AVA for — to replace a real human behind a desk making calls. Your virtual agents should be treated as trained staff, not static advertising tools. Configure them the same way you’d brief a new hire: give them clear goals, focused messaging, and the context they need to adapt. While selling your product is important, the AI works best when it earns attention through a genuine conversation. Its strength lies in adapting to client cues and delivering the right message at the right time — not overwhelming leads with one-sided pitches.
Also, avoid scripting client or representative names like “John” or “Daniel.” Use custom values such as {{client_firstname}} and {{representative_name}} to make scripts reusable and scalable.
❌ What not to do:
“Hi John! This is Ava calling from GrowthPartners Inc., and I work closely with Daniel our Senior Marketing Director. I’m calling because we wanted to offer you a chance to try our full package, which includes AI reporting, CRM integration, onboarding support, and a ton more. So are you interested today or should I follow up tomorrow?”
✅ Better approach:
“Hi client_firstname, this is Ava from company_name. I just wanted to follow up and see if you’d like to continue the conversation about our offer.”
Any additional service or feature info should be stored in Key Information, so the agent only references it if the client brings it up.
2. Structure Informational Fields Clearly
Fields like Key Information, are meant to store factual support content, not prompts, not scripts, and definitely not paragraphs full of filler. These fields are accessed conditionally by the AI, so clarity and clean formatting are critical.
Also, don’t include instructions or conversational phrasing in these fields. Avoid dumping disorganized thoughts with no punctuation, and absolutely avoid scripting names.
❌ What not to do:
“Tell John that our Standard Plan has unlimited calls and the pro plan is $349/month and Daniel gives 10% discounts to referrals and just say the analytics are good and also if they mention price tell them that the discount only applies if they pay yearly and say that right after.”
✅ Better formatting:
- Standard Plan – Unlimited calls
- Pro Plan – $349/mo
- Includes – AI analytics, onboarding support
- Discount – 10% for annual billing
- Representative – representative_name
Proper formatting keeps the field easy to parse and avoids errors during live responses.
3. Know When Scripting + Prompting Can Be Mixed
Some fields — such as Objection Handling — work well with a mix of both scripting and prompting. In these cases, you can give a general instruction (the prompt), followed by a sample line (the script). Just remember to keep it flexible and scalable using custom values.
🎯 Example of valid mixed scripting + prompting:
- Prompt: “If the client says it’s too expensive, explain the long-term ROI or offer a shorter free option.”
- Script: “Totally understandable, client_firstname. We also offer a free 15-minute strategy call to help you evaluate fit. Would that help?”
Final Note
Use each field for a streamlined purpose — and always pair it with custom values like {{client_firstname}} or {{representative_name}} to keep dialogue dynamic and scalable. Well-structured input reduces awkward pauses, interruptions, and confusion — making your scenarios more efficient and professional.
FAQs & Troubleshooting
General Questions
Configuration
For additional questions or guidance, try using our Virtual Support Agent! Available 24/7 to help resolve most issues quickly at thinkrr.ai/support.
If you still need assistance, visit our support site at help.thinkrr.ai and submit a Ticket or contact our team directly at hello@thinkrr.ai.