# Channels — How customers use the agent There are two main ways customers can interact with the CS Agent today. Your integration plan should choose **one primary path** for production, and use the other for testing or backup. --- ## 1. LiveChat (production-style path) **What it is:** The same LiveChat product many teams use for website chat. The agent connects to LiveChat via **API** so that when a customer sends a message in the widget, the **agent** can reply automatically. **What the customer sees:** - The normal chat widget on your site. - On the IKON168 Customer Portal (`https://portal.ikon168.com`), customers can click either the floating widget or the `Chat Support` / `Open LiveChat` buttons to open the popup. - Messages that look like **plain text** from a support agent (no markdown formatting in the chat UI). - If the system **hands off to a human**, the customer may see a short message that someone will join them; your **human agent** then continues in the **LiveChat agent dashboard** (same conversation thread). **What your team needs:** - A LiveChat **license** and the **widget** installed on the site where you want chat. - DNS for `portal.ikon168.com` if using the official IKON168 Customer Portal for UAT or production-style validation. - **API credentials** (account ID and Personal Access Token) configured on the server that runs the CS Agent — these are supplied by your operations team and stored securely (not in this document). **Integration note:** The agent **polls** LiveChat for new conversations. Your LiveChat setup and routing rules should still allow chats to reach agents when the AI escalates or when the kill switch is turned on. --- ## 2. Web demo (UAT / internal testing) **What it is:** A **browser page** served by the same application. It is a **simplified chat** for demos and training. **Production URL (after DNS and HTTPS):** `https://agent.ikon168.com` — same UI with TLS (e.g. Caddy on the deployment host). For local development: `http://localhost:8080`. For cloud UAT before HTTPS: `http://:8080` as documented in deployment configuration. **What the customer sees:** - A simple chat page with a text box and send button. - Optional **test messages** in the sidebar to try common scenarios quickly. **When to use it:** - **User acceptance testing (UAT)** before go-live. - **Training** for staff who want to see how the AI answers without using the live widget. - **Not** a substitute for a public marketing website unless you intentionally expose it — **protect it** in production (password on the dashboard, HTTPS, network restrictions). --- ## 3. End-to-end flow (simple) ``` Customer opens chat (LiveChat widget / web demo) → Customer sends a message → CS Agent classifies intent and answers (or escalates) → If escalation: Telegram alert to staff + human continues in LiveChat when applicable ``` --- ## 4. SuperBO (backoffice) **What it is:** Your **SuperBO** web console is used for **backoffice operations** (player lookup, account actions, etc.). The CS Agent project has **automation** capability for the browser (RPA) for future workflows, but **day-to-day player chat** is still through **LiveChat** (or the web demo for tests). **For your team:** Continue to use SuperBO as you do today for account-level work. **Chat** with the player remains **LiveChat** when the AI hands off or when the kill switch routes all traffic to humans. --- Next: [03-Your-Team-Escalations-and-Telegram.md](03-Your-Team-Escalations-and-Telegram.md)