Model & behavior settings
The AI Configuration section controls which model answers your visitors and how it behaves. Open
Dashboard → Settings🖼️ [Image] — The AI Configuration card showing the Model, Temperature, Max Tokens and Max History fields.
AI Configuration
| Field | Options / range | Default | What it does |
|---|---|---|---|
| Model | text text text | text | The model that writes the final answer. text text text |
| Temperature | text text text | text | Creativity vs. consistency. Low ( text text |
| Max Tokens | text text | text | Maximum length of a single reply. Higher values allow longer answers but cost more. |
| Max History (messages) | text text | text | How many previous turns the bot remembers within the current chat. Higher values keep more context but add tokens to every request. |
The Model field selects the model used to write the answer. Some internal steps (intent routing, planning) always run on a fast small model regardless of this setting — see Multi-step reasoning and Intent routing.
Widget enabled
In the Widget section, the Widget enabled toggle controls whether your chat widget loads on your site. When it's off you'll see the warning "Your widget is currently hidden from all visitors." The embed script still loads but renders nothing — useful for pausing the bot without removing code. See Embedding the widget.
🖼️ [Image] — The Widget section with the "Widget enabled" toggle and the hidden-widget warning banner.
Tuning tips
- Support / factual bots: Temperature , Max Tokenstext
0.2–0.5.text512–1024 - Conversational / sales bots: Temperature .text
0.6–0.8 - Raise Max History only if visitors reference earlier messages a lot; it directly increases per-message cost.
- Keep Max Tokens modest — most good answers fit in under 1024 tokens.
Click Save Changes to apply. The System Prompt is required before you can save.