This page documents every setting available on a ChatbotIQ bot. Access bot settings by clicking Edit on a bot, or the gear icon (⚙) in the Playground header.
| Setting | Type | Default | Description |
|---|
| Name | Text | (required) | The display name for your bot (e.g., “Support Bot”). Shown in the dashboard and conversation list. |
| Description | Text | (empty) | Optional description for your reference. Not shown to visitors. |
| Active | Toggle | On | Whether the bot accepts messages. When off, the widget won’t load and API calls are rejected. |
| Setting | Type | Default | Description |
|---|
| LLM Provider | Select | OpenAI | The AI provider. Options: OpenAI, Mistral, Anthropic, Google Gemini. |
| LLM Model | Select | GPT-5 Mini | The specific model to use. Available models depend on the selected provider. |
Embedding Model is not a standard per-bot setting. It is managed by administrators/support as an Embedding Model Override on the workspace admin page; the default is text-embedding-3-large.
| Setting | Type | Default | Description |
|---|
| System Prompt | Text | (empty) | Your bot’s personality, tone and scope, layered on top of ChatbotIQ’s built-in instructions. The built-in rules (answer from your knowledge base, cite sources, say when something is not covered) always take precedence, so the system prompt cannot switch them off; use the toggles below for that. Also called “base prompt.” |
| Language | Select | Auto-detect | The language your bot responds in. Defaults to Auto-detect, which mirrors the language of each user message (falling back to English if unclear). See Supported Languages for the full list. |
| Temperature | Number (0-1.0) | (model default) | Controls response randomness. Lower = more consistent, higher = more creative. Leave blank to use the model’s default. Only shown for Mistral models; hidden for OpenAI, Anthropic and Gemini models (reasoning models that reject a temperature setting). |
| Response Verbosity | Select | (model default) | How long responses should be: Low (a few sentences, no extras), Medium (balanced), or High (detailed, with the examples and edge cases your content has). |
| Allow General Knowledge | Toggle | Off | When off, every fact comes from your knowledge base and the bot says so when a question is not covered. When on, the bot may add brief background or definitions, marked as general knowledge, and may briefly answer a general question your content does not cover; specific facts, figures and procedures still come only from your content. |
| Content Type | Select | (auto-detected) | The kind of content your bot serves, which sets its tone and type-specific rules: technical docs (step-by-step and API answers, code examples only from your content), e-commerce (product listings with price and availability when present), customer support (next steps, no promises about individual cases), legal (wording close to the source and a single “not legal advice” note), education (explanations with the examples in your material). Also affects output filtering. |
| Setting | Type | Default | Description |
|---|
| Sources | Multi-select | All sources | Which knowledge sources the bot draws from. Choose specific sources or “All sources” to use everything in your knowledge base. |
| RAG Score Threshold | Number (0-1) | (auto-tuned) | Controls how strictly the bot matches your content. The slider sits at 0.25 by default, but leaving it there stores no value: the retrieval engine then applies its own tuned default (around 0.30), which also varies by content type (lower, ~0.20, for e-commerce; higher, ~0.40, for legal). Raising the value makes the bot stricter (only answers when very confident); lowering it makes the bot more lenient (answers more questions but may be less precise). |
| Setting | Type | Default | Description |
|---|
| Suggested Questions | List | (empty) | Pre-written prompts shown when the chat opens. Helps guide users to common topics. |
| Show Suggested Questions | Toggle | On | Whether to display suggested questions. |
| Suggested Questions Count | Number | 3 | How many suggested questions to show at once. |
| Include Source Citations | Toggle | On | Show citation links at the bottom of responses, linking to the source pages used. |
| Inline Citations | Toggle | Off | Show numbered citation superscripts within the response text (in addition to footer citations). |
| Include Images | Toggle | Off | Allow the bot to include images from your knowledge base in responses. |
| Allow User Images | Toggle | Off | Let users upload images in the chat (for multimodal AI models). |
| Include Links | Toggle | On | Include hyperlinks in bot responses. |
| Glossary | Key-value pairs | (empty) | Custom term definitions. When the bot encounters these terms, it uses your definitions. Add terms via the dashboard - no coding needed. |
| Warning Message | Text | (empty) | A disclaimer banner pinned at the top of the chat window, shown to every visitor for the whole conversation. Commonly used for an AI accuracy notice. Supports Markdown, including links. Leave empty to show no banner. |
| Setting | Type | Default | Description |
|---|
| Privacy Mode | Select | Private | Public = anyone can chat (required for website widget). Private = only workspace members. Authenticated = any logged-in user. |
| Allowed Domains | List | (empty = all) | Domains where the widget is allowed to load. Empty means any domain. Supports wildcards (e.g., *.example.com). |
| Record IP Addresses | Toggle | Off | Store visitor IP addresses with conversations. Useful for analytics but may have privacy implications. |
| PII Redaction | Toggle | Off | Automatically detect and mask personally identifiable information (emails, phone numbers, etc.) in chat messages. See Enable PII Redaction. |
| Setting | Type | Default | Description |
|---|
| Messages per Minute | Number | (plan default) | Maximum messages this bot answers per minute in total, across all visitors combined. Empty means your plan limit applies. |
| Messages per Hour | Number | (plan default) | Maximum messages this bot answers per hour in total, across all visitors combined. |
When a limit is reached, the visitor sees a standard “rate limit exceeded, please try again later” notice. This text is fixed and cannot currently be customised.
Per-visitor limits are always on. Independently of the settings above, every public bot enforces a built-in limit of 15 messages per minute and 120 messages per hour per visitor IP address. This applies automatically and cannot be disabled.
The bot-level settings above are a total ceiling across all visitors, not a per-visitor limit. Three layers apply together (per visitor IP, per bot, and your workspace plan limit), and whichever is reached first takes effect. Rate-limited requests are rejected before the AI model runs, so they consume no credits.
| Setting | Type | Default | Description |
|---|
| Web Search Enabled | Toggle | Off | When on, the bot performs a parallel web search scoped to your source domains during retrieval. Supplements knowledge base results with fresh web content. See Enable Web Search. |
| Setting | Type | Default | Description |
|---|
| MCP Enabled | Toggle | Off | Expose this bot as an MCP server so AI tools can query your knowledge base. Requires a paid plan (Personal or higher). Choose between API Key (authenticated) or Public mode when enabling. See Connect via MCP. |
| MCP API Key | Generated | (none) | API key for authenticating MCP connections. Generated when you enable MCP in API Key mode. Shown with a prefix for identification; the full key is shown once at creation. |
| Show MCP Tools in Widget | Toggle | Off | Display available MCP tools in the chat widget. Only available when MCP is enabled in Public mode, generating an API key force-disables this toggle. |
| Setting | Type | Default | Description |
|---|
| Handoff Enabled | Toggle | Off | Allow conversations to be transferred to a live agent. See Enable Human Handoff. |
| Handoff Trigger | Select | Button | When handoff is offered: Button = user clicks a button, Auto = bot decides automatically, Both = either method. |
| Offline Message | Text | (empty) | Message shown when no agents are available (e.g., “Our team is offline. We’ll get back to you soon.”). |
| Agent Label | Text | (empty) | Display name for the human agent in the chat (e.g., “Support Agent”). |
| Agents Available | Toggle | Off | Whether live agents are currently online and ready to respond. |
| Setting | Type | Default | Description |
|---|
| Contact Form Enabled | Toggle | Off | Show an inline contact form in the chat when the bot can’t help. Collects name, email, and message, delivered to your inbox. |
| Contact Form Trigger | Select | Button | When the form is offered: Button = a “Contact us” button in the widget, Auto = shown when the bot can’t answer, Both = button + auto-detect. |
| Button Display | Select | After response | When the contact form button is visible: After response = only after a bot response, Always = always visible in the widget. |
| Custom Fields | JSON | (default: name, email, message) | Define custom form fields. Each field has a name, label, type (text, email, or textarea), and whether it’s required. Leave empty to use the default fields. |
| Notification Email | Text | (empty) | Override email for form submissions. If empty, notifications go to all workspace members. |
| Success Message | Text | (default) | Message shown after the user submits the form. Leave empty for the default message. |
