> ## Documentation Index > Fetch the complete documentation index at: https://docs.cura.so/llms.txt > Use this file to discover all available pages before exploring further. # How your AI agent works > Your AI agent is a tireless front-desk assistant that reads every incoming message, answers from what you've taught it, and helps customers book — all in seconds, on every channel. Your AI agent is a tireless front-desk assistant that reads every incoming message, answers from what you've taught it, and helps customers book — all in seconds, on every channel. This article explains what it does behind the scenes, so you know what to expect and how to shape it. In short: the AI listens, understands, replies from *your* business knowledge, and hands off to a human the moment it should. *** ## The capture → answer → book loop Every conversation follows the same simple loop. * **Capture.** When a message arrives on any connected channel (LINE, WhatsApp, Instagram, Messenger, Telegram, or your website), the AI reads it along with the recent history of that conversation. * **Answer.** It works out what the customer wants and replies in their language, using your business knowledge. * **Book.** When the customer is ready, the AI can check real-time availability and create a **booking request**. New bookings are *pending* until your team approves them, so the AI never tells a customer they're "confirmed" — it captures the request and your team gives the final yes. ```mermaid theme={null} flowchart LR M(["New message
on any channel"]) --> C["Capture
read the message
and recent history"] C --> A["Answer
reply in the customer's
language, from your knowledge"] A --> B["Book
create a pending
booking request"] B --> Y["Your team gives
the final yes"] A -. "when it needs a person" .-> H["Hand off
to your team"] ``` ## It answers from your knowledge, not guesswork The AI is *grounded* — it builds answers from the knowledge you give it, rather than inventing details. It draws on three sources: * **Your Knowledge Base** — uploaded documents (PDF, Word, text, Markdown) covering policies, treatments, and aftercare. * **Your FAQs** — exact question-and-answer pairs for the things customers ask most. * **Your Services** — each treatment with its description, price, and any promo price, so quotes are accurate. The richer these are, the more confident and correct your AI becomes. See [Build your Knowledge Base](/ai-agent/build-your-knowledge-base) to get started. ## It sounds like your business You control how the AI speaks. **Rules** set its **tone** (friendly, neutral, matter-of-fact, professional, or humorous) and **length** (short, standard, or long). For finer control, **Advanced AI instructions** let you describe communication style, when to escalate, and any other guidance — in your own words. The AI Agent settings in Cura where you set the agent's tone, length, and rules ## A human is always one step away The AI is built to know its limits. When a customer asks for a person, when it can't help, or when something feels sensitive, it **escalates** — handing the conversation to a human on your team. You can also **take over** any conversation yourself at any time. This is the "human-in-the-loop" safety net: the AI handles the routine, your team handles the rest. ## You stay in control * **Master AI switch** turns the AI on or off for an inbox. * **AI schedule** lets it reply only during the hours you choose. * **Take over** any conversation at any time to reply yourself. Think of the AI like a well-trained new hire. The clearer your "training" — Knowledge Base, FAQs, Services, and Rules — the better it performs. Start simple, then refine as you watch real conversations. *** ### Related articles * [Build your Knowledge Base](/ai-agent/build-your-knowledge-base) * [Setting your AI's tone and length (Rules)](/ai-agent/ai-tone-and-length) * [Escalation: when the AI hands off to a human](/ai-agent/escalation-handoff) * [The Master AI switch and per-conversation control](/ai-agent/master-ai-switch)