> ## Documentation Index > Fetch the complete documentation index at: https://www.revve.ai/docs/llms.txt > Use this file to discover all available pages before exploring further. # What Is a Voice Agent? > Learn what Voice Agents are, how they handle real phone calls, and when to use one instead of (or alongside) a Chat Agent. A **Voice Agent** is an AI agent that talks to your customers on real phone calls — answering inbound calls on your phone number, or making outbound calls for follow-ups, qualification, reminders, and campaigns. It speaks and listens in natural conversation: callers can interrupt it, pause to think, and ask anything, just like talking to a person. Image List of all voice agents ## How a call works On every conversational turn, the Voice Agent runs a three-stage pipeline in real time: 1. **Speech Recognition (STT)** transcribes what the caller says as they speak. 2. **The AI model (LLM)** decides what to say next, following your instructions, your knowledge base, and the conversation so far. It can also act — transfer the call, look up information, or trigger a tool. 3. **Voice Synthesis (TTS)** speaks the response in the voice you selected. You choose the provider, model, and language for each stage in [Model Settings](/docs/voice-agents/voice-agent-settings), and you tune the conversational rhythm — when the agent considers the caller "done talking", how it handles interruptions, what it does during silence — in Advanced Settings. The defaults work well out of the box; tuning matters most for noisy environments and specific languages. ## What Voice Agents can do | Capability | What it means in practice | | ------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Inbound calls** | Assign a phone number to the agent and it answers every call — a receptionist that never misses one. | | **Outbound calls** | Trigger calls from [Campaigns](/docs/campaigns/campaign-overview), the API, or manually — follow-ups, reminders, qualification calls. | | **Call transfer** | Hand the call to a human in one of two modes: bridge transfer (the AI stays on the line and keeps logging) or cold transfer (full handoff). | | **Voicemail handling** | Detect answering machines and either hang up or leave a templated message. | | **Knowledge base answers** | Attach the same knowledge bases your Chat Agents use, so the agent answers from your content. | | **Dynamic variables** | Personalize calls with contact fields like `{{fullName}}` or `{{companyName}}` injected at call time. | | **Recording, transcripts & analysis** | Every call is recorded and transcribed; an AI summary, sentiment, and extracted data fields appear in Call History. | | **Tools & webhooks** | Call external APIs mid-conversation, send SMS, check and book calendar appointments (Calendly, Cal.com, Chili Piper), and notify your systems when calls end. | ## Two engine types Like Chat Agents, every Voice Agent uses one of two engines, chosen at creation: * **Single Prompt** — one system prompt drives the whole call. Fastest to set up, most natural-feeling; right for receptionists, support lines, and open-ended conversations. * **Conversational Flow** — a visual graph of nodes with explicit transitions. Right when the call must follow a strict sequence: verify identity → present offer → capture decision → wrap up. If you're unsure, start with Single Prompt. The decision logic is the same as for chat — see [Understanding Engine Types](/docs/getting-started/understanding-engine-types). ## Where everything lives Once inside a Voice Agent, you'll work across these tabs: | Tab | Purpose | | --------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Dashboard** | Call volume, outcomes, and performance for this agent. | | **General** | Name, description, region, and the system prompt. | | **Model Settings** | Speech Recognition, AI Model, and Voice Synthesis configuration. | | **Knowledge Base** | Attach content sources the agent answers from. | | **Advanced Settings** | Call behavior, conversation rhythm, voicemail, phone number, call transfer, tools, webhooks, and analysis. | | **Evaluation** | Define a grading policy that automatically scores every completed call against a weighted set of metrics. See [Voice Agent Evaluation](/docs/voice-agents/voice-agent-evaluation). | | **Preview** | Talk to the agent from your browser, or generate a test phone call link. | | **Versions** | Draft and publish — changes never go live until you publish. Past versions are viewable read-only; there's no rollback. See [Voice Agent Versioning](/docs/voice-agents/voice-agent-versioning). | ## Voice Agent or Chat Agent? They're complementary, and most teams run both: * **Chat Agent** — visitors come to you (website, WhatsApp, email). Best for support volume and capturing intent at the moment of interest. * **Voice Agent** — you reach out, or callers dial in. Best for leads that need a conversation, time-sensitive follow-up (speed-to-lead), and audiences that simply prefer phone. A common pattern: a Chat Agent qualifies a website visitor and captures their number, then a [Campaign](/docs/campaigns/campaign-overview) has a Voice Agent call them within minutes. ## What's Next * [Creating Your First Voice Agent](/docs/voice-agents/creating-your-first-voice-agent) — from template to a live test call in a few minutes. * [Voice Agent Settings](/docs/voice-agents/voice-agent-settings) — every setting, its default, and when to change it. * [Voice Agent Evaluation](/docs/voice-agents/voice-agent-evaluation) — automatically grade completed calls against a weighted set of metrics. * [Build a Voice Agent That Books Appointments Mid-Call](/docs/guides/appointment-booking-voice-agent) — a complete recipe for the calendar-booking tools mentioned above.