How the AI Chatbot and Live Chat Work
The relationship between the AI Support Chatbot and Live Chat, how they form a two-layer support pipeline, and where to configure each.
Overview
Plenix has two separate but connected chat systems that together form a two-layer support pipeline:
- AI Chatbot (Autonomy → Chatbot) — automated first-line deflection using the Knowledge Base
- Live Chat (Inbox → Live Chat) — real-time human agent conversations
Understanding how they differ — and how they hand off to each other — is essential for configuring your support channel correctly.
Layer 1: AI Support Chatbot
Where it lives: Autonomy → Chatbot What it does: A widget you embed on your website or client portal. When a visitor types a question, the chatbot:
- Searches your published Knowledge Base articles
- Uses AI to synthesise a concise answer from matching articles
- Returns the answer instantly — no human involved
If the chatbot cannot find a relevant KB article, it falls back to one of three options (configurable in the Chatbot settings):
- Auto-create a support ticket — the visitor's question becomes a ticket; they're told a reference number
- Escalate to Live Chat — the conversation is handed off to a human agent (see Layer 2)
- Show a generic message — "We couldn't find an answer, please contact support"
Embed code: The Chatbot page gives you a code snippet to paste into the </body> tag of any webpage to show the widget.
Conversations tab: All chatbot sessions (visitor messages + bot responses) are logged and visible in the Conversations tab on the Chatbot page — including whether a ticket was created.
Layer 2: Live Chat (Human Agents)
Where it lives: Inbox → Live Chat What it does: A real-time chat inbox for your support team. Staff see incoming chat sessions and can reply in real time. Visitors interact via a separate embeddable widget.
When does a session appear here?
- When the AI Chatbot escalates (because it couldn't answer and fallback = "live chat")
- When a visitor starts a chat via the Live Chat widget directly (embedded separately from the chatbot)
Status flow: Sessions start as Waiting (no agent has replied yet) → go Active when an agent joins → become Closed when the agent ends the session.
Embed code: The Live Chat page has its own embed snippet — go to Inbox → Live Chat and click Embed Widget in the top right.
How they work together
Visitor asks a question
│
▼
AI Chatbot checks KB
│
┌────┴────┐
│ Found │ → AI synthesises answer → Visitor satisfied → Done
└─────────┘
│
┌────┴────┐
│ Not │
│ found │ → Fallback setting
└─────────┘
│
┌────┼────────────────────────┐
▼ ▼ ▼
Ticket Live Chat escalation Generic message
created (Waiting in Live Chat)
Setting up the chatbot correctly
- Go to Autonomy → Chatbot
- Enable the chatbot (toggle)
- Set the bot name and greeting message
- Choose the fallback: "Escalate to live chat" if you want human backup; "Auto-create ticket" if you want zero-touch
- Publish KB articles (in Service Desk → Knowledge Base) — the chatbot is only as good as your KB content
- Grab the embed code and add it to your website
Setting up Live Chat
- Go to Inbox → Live Chat
- Click Embed Widget to get the embed snippet
- Paste the snippet on your website (or portal — it works on both)
- Incoming chats will appear in this inbox; assign agents to monitor it during support hours
Key differences
| AI Chatbot | Live Chat | |
|---|---|---|
| Who responds? | AI (automated) | Human agents |
| Where configured? | Autonomy → Chatbot | Inbox → Live Chat |
| Data stored? | chatbotSession (Q&A logs) | Visitor chat sessions |
| Can create tickets? | Yes (on fallback) | No (manual if needed) |
| Available 24/7? | Yes | Only when agents are online |
Where to go next
Was this article helpful?