Chatbot Dashboard: Tabs & Settings
Chatbot Dashboard
Open Dashboard and select an AI assistant in the sidebar. Its workspace is organized into five main tabs:
| Tab | What you manage |
|---|---|
| Overview | Readiness, AI connection, number of knowledge sources, and available channels |
| Settings | Names, allowed websites, language, timezone, AI provider, model, answer variation, and role |
| Knowledge | Websites, documents, text, cloud folders, and other content used in answers |
| Functions | Live chat, question collection, leads, feedback, appointments, and URL search |
| Channels | Chat widget, inline chat, AI Team Wiki, WhatsApp, phone, and email |
Advanced settings sits below the main tabs and contains MCP Server Configuration and API Connectors. There is no separate Integration tab. Website installation is part of Channels → Chat widget.
On small screens, the same sections appear in a section selector instead of a horizontal tab bar.
Header actions
The header applies to the currently selected assistant:
- Test chatbot opens the real chat preview so you can verify answers.
- Chatbot actions → View setup opens the guided setup again.
- Chatbot actions → Copy chatbot ID copies the ID used by widgets, APIs, plugins, and support.
- Chatbot actions → Delete chatbot permanently removes the assistant and its data after confirmation.
The status beside the assistant name shows whether the assistant is ready or still needs setup.
Overview
Overview is a quick readiness check. It shows:
- whether an AI provider is connected;
- how many knowledge sources the assistant has;
- how many channels are available;
- direct actions to connect AI, add knowledge, test the assistant, or configure a channel.
Use this tab after creating an assistant to see what is still missing. Detailed editing happens in the other tabs.
Settings
The Settings tab controls identity, access, language, and answer behavior. A save indicator at the top reports saved, saving, or failed changes.
Name & Access
| Setting | What it changes |
|---|---|
| Service Name / Company Name | The company or brand the assistant represents. It can be used in answers and shown to visitors. |
| Chatbot name | The internal name shown in your dashboard and assistant list. Visitors do not see it. |
| Allowed websites | Domains permitted to load this assistant's widget. Add every production domain and subdomain that should use it. |
| Assigned user | White-label organization admins can transfer the assistant to an existing customer account. The selection saves immediately and the assigned customer's quota applies. |
Enter domains without paths, for example example.com or help.example.com. If the widget does not load, check this list first.
For the complete transfer workflow, including which chatbot data and usage move to the customer, how to take the chatbot back, and the external API-key caveat, see Assigning an admin-created chatbot to a customer.
Language & Time Zone
| Setting | What it changes |
|---|---|
| Timezone | Used for appointment scheduling and time references in answers. |
| Default Language | Fallback language for short or ambiguous messages such as “Hi”. This is especially relevant for messaging channels. |
The assistant can still answer in other languages when it can detect the visitor's language. The default resolves cases where there is not enough text to detect one reliably.
Advanced AI settings
Expand Advanced AI settings under Answer behavior to configure:
| Setting | What it changes |
|---|---|
| AI provider | The service that generates answers. Provider options show included quota and processing region where available. |
| Model | The answer model. The list shows quality, speed, region, and message multiplier where applicable. |
| Answer creativity | Lower values produce more consistent answers; higher values allow more variation. |
| Show answer word by word | Streams the response while it is generated instead of waiting for the complete answer. |
The model list changes as providers release new versions. Treat the dropdown in the dashboard as authoritative.
Model multipliers affect the monthly allowance. If a model is marked 6×, both the visitor message and the answer count at that multiplier. The dashboard shows the resulting usage before you select it.
Answer style and role
Open Answer style and role to define the assistant's standing instructions. You can:
- choose a preset such as website support, shopping assistant, lead generation, onboarding, booking, or team wiki;
- generate a draft from existing knowledge sources;
- write or edit the role yourself;
- set tone, scope, boundaries, escalation rules, and preferred answer format.
A useful role is specific but short. State who the assistant represents, what it should help with, what it must not invent, and when it should hand over to a person.
You are the support assistant for ACME. Answer questions about products,
shipping, and returns from the connected knowledge sources. Keep answers
concise. If the sources do not contain the answer, say so and offer human help.
Never invent prices, availability, or policy details.
Test the same set of representative questions after changing the provider, model, creativity, or role. This makes quality changes easier to spot.
Knowledge
The Knowledge tab contains everything the assistant may use as grounded information. It shows account-wide character usage plus a searchable, filterable source table with type, name, category, indexing status, update time, and actions.
From here you can:
- add websites, documents, text, Confluence, Notion, SharePoint/OneDrive, Google Drive, Nextcloud/WebDAV, or SFTP/FTPS;
- search and filter sources by type or category;
- re-import one source or all affected sources;
- edit, inspect, download, stop, or delete a source where supported;
- select multiple sources for bulk re-indexing or deletion.
See Knowledge Sources for source-specific fields, sync intervals, exclusions, and re-indexing.
Functions
The Functions tab adds capabilities to the selected assistant. It also warns that enabling more functions can increase response time.
| Function | Main settings |
|---|---|
| Live Chat | Human takeover, request notifications, recipients, and timeout before the bot resumes |
| Questions | Collect unanswered questions, optionally log answerable questions, and send notifications |
| Leads | Collect contact information and interests, with optional notifications |
| Feedback | Let visitors rate answers and leave comments, with optional notifications |
| Appointments | Calendar provider, meeting title, duration, lead time, buffer, timezone, working hours, phone-call access, and notifications |
| URL Search | Search configured external URLs in real time or analyze a URL mentioned by the visitor |
Each enabled function links to its dashboard where you review the collected chats, questions, leads, feedback, or bookings. See Functions Overview for detailed setup.
Channels
Channels define where people can use the selected assistant. The available channel cards are:
| Channel | Purpose |
|---|---|
| Chat widget | Floating website button and chat window |
| Inline chat | Chat embedded directly inside a page layout |
| AI Team Wiki | Private or controlled knowledge portal on a WebChatAgent subdomain |
| Messages through a connected WhatsApp Business account | |
| Phone (Beta) | Incoming calls using the same knowledge base and approved tools |
| Email (Beta) | Mailbox replies saved as drafts or sent directly |
Chat widget
Open Channels → Chat widget. The installation panel lets you copy the pre-filled code, check whether the widget is loading on an allowed domain, and open platform guides for WordPress, Shopify, Wix, Squarespace, and Webflow.
Widget settings are split into five tabs:
| Widget tab | Settings |
|---|---|
| Appearance & content | Theme, width, height, title, font, colors, avatar, button image, welcome image, welcome message, speech bubble, and privacy notice |
| Behavior | Disable widget, initially open, microphone, sources, AI label, translation, file upload, notification sound, WhatsApp button, hidden URLs, and desktop/mobile offsets |
| Automatic engagement | Proactive triggers, timing and behavior rules, frequency controls, templates, analytics, and optional consent integration |
| Buttons & forms | AI-generated or manual prompt buttons, Action Bar items, and custom forms |
| Advanced | Language detection source, branding, custom CSS, presets, and stable selector reference |
The AI label under bot replies is always on because it is required for transparency. Use Test Your Chatbot for a real preview and Save Changes to publish widget changes.
See Chat Widget Settings, Proactive Triggers, Action Bar & Forms, and Custom CSS.
Inline chat
Channels → Inline chat has its own installation code and independent settings:
- Appearance: title, theme color, font, height, and welcome message;
- Images: optional welcome image by upload or URL;
- Content & Behavior: welcome message, sources, mandatory AI label, automatic translation, file upload, and prompt buttons;
- Advanced: language detection, branding, and custom CSS.
The live preview uses the real assistant when you send a message. Human takeover is not available in inline chat. See Inline Chat.
Other channels
- AI Team Wiki: create and manage the assistant's wiki subdomain. Live chat and human takeover are not available there.
- WhatsApp: connect a WhatsApp Business account through Meta's authorization flow.
- Phone: assign a number, set voice and persona, choose call tools, limits, transfers, transcript consent, and test calls.
- Email: enter IMAP/SMTP details, test the connection, map folders, choose draft or direct-send behavior, set salutation, history use, instructions, signature, authorization, and channel status.
See AI Team Wiki, WhatsApp, Phone, and Email.
Advanced settings
Expand Advanced settings below the main tabs when the assistant needs external services:
- MCP Server Configuration connects Model Context Protocol servers and lets you control their individual tools.
- API Connectors connects REST endpoints for live data such as order status, ticket details, or availability.
These are advanced connections, not channels and not items in the Functions tab. See MCP Servers and API Connectors.
Safe change workflow
- Confirm the selected assistant in the heading and sidebar.
- Change one group of settings at a time.
- Wait for the saved state or use the section's Save Changes button.
- Re-index knowledge if a provider change requires it.
- Use Test chatbot with repeatable questions.
- For website changes, test both desktop and mobile on an allowed domain.