Conversational AI Overview
Voxvaani’s Conversational AI features enable intelligent, automated conversations across voice and chat channels using AI-powered agents and chatbots. The AI engine is fully managed by Voxvaani — no separate AI provider accounts, API keys, or model configuration are needed on your side.
AI Agents vs Chatbot
| Capability | AI Agents (Voice) | Chatbot (Text) |
|---|---|---|
| Channel | Phone calls (integrated into IVR flows) | Text-based (WhatsApp, web chat) |
| Interaction | Real-time voice conversation with text-to-speech output | Asynchronous or real-time text messaging |
| Primary Use Cases | Customer support hotlines, lead qualification calls, FAQ automation, appointment scheduling | Customer inquiries, lead capture, support ticket creation, FAQ responses |
| Flow Control | Added as a node in IVR Flow Builder; can transfer to human agents | Handles the full conversation or escalates to a human agent |
When to Use AI Agents
- You receive a high volume of repetitive phone inquiries
- You need 24/7 phone support without staffing night shifts
- You want to qualify leads by phone before routing to a sales team
- You have an IVR menu and want to replace rigid key-press options with natural conversation
When to Use Chatbot
- Your customers prefer text-based communication (WhatsApp, web chat)
- You need to capture leads from inbound messages automatically
- You want to reduce support ticket volume by answering FAQ autonomously
- You are running WhatsApp campaigns and need to handle responses
Quick Start
- Create your first Agent or Bot:
- For voice: Navigate to AI Agents, define a system prompt, pick the transcription language and TTS voice, and add the agent as a node in your IVR Flow Builder.
- For text: Set up a Chatbot, configure its prompt and behavior, then connect it to your WhatsApp or web chat channel.
- Test the conversation — use the built-in test console to try realistic caller messages before going live.
How It Works
Detailed Flow
- Input Reception — A call reaches an AI Agent node in the IVR, or a message arrives at the Chatbot
- Context Loading — The system loads the configured system prompt, conversation history, and contact context
- LLM Processing — The message or transcribed speech is sent to the platform’s AI engine along with the system prompt
- Intent Recognition — The model determines the user’s intent (e.g., “check balance”, “speak to agent”, “create ticket”)
- Action Execution — Based on intent, the system either:
- Generates a text/speech response
- Creates a support ticket
- Captures and saves lead data
- Transfers to a human agent or fallback node
- Response Delivery — For voice: TTS is played to the caller. For chat: text response is sent back.
Requirements
- A properly configured AI Agent or Chatbot
- For voice agents: a phone number connected to your IVR flow