VEX Settings | Yaplet

The VEX settings page is organized into five tabs: General, AI Identity, Agent Handoff, Advanced, and Danger Zone.

General

Basic chatbot configuration and connected knowledge sources.

Chatbot Name: A friendly name to identify this chatbot in your dashboard.
Enable Chatbot: Turn the chatbot on or off. When disabled, the AI won't respond to visitors.

Widgets & Knowledge Sources

A table showing which widgets use this chatbot and what knowledge sources are attached to each one. The AI uses knowledge bases and documentation as its source of truth when answering questions.

The table shows:

Widget: The connected widget (links to its settings)
Knowledge Base: The knowledge base providing context, if configured
Documentation: The documentation providing context, if configured

Each knowledge base and documentation entry has a View chunks button that lets you preview how the content was processed and split into chunks for the AI.

When you set up a knowledge base or documentation for a widget, the AI is automatically trained on all the information it contains. If you later edit or update the content, changes are automatically synchronized with the AI — no manual retraining needed.

Widget-to-chatbot connections are managed in the widget settings.

AI Identity

Tell the AI about your company so it can give accurate, on-brand responses.

Company Name: The AI uses this name when referring to your company in conversations.
Website URL: Your company's website. The AI may reference this when helping visitors.
Emoji Usage: By default the AI uses emojis for a friendly tone. Disable this for a more formal style.
Custom Instructions: Extra instructions sent with every AI prompt. Use this to define tone, rules, or specific information the AI should always know (e.g. "Always greet users by name. Never discuss competitor products. Our return policy is 30 days.").

When the AI Doesn't Know the Answer

Choose what happens when the AI can't find relevant information in your knowledge base to answer a question:

Option	Behavior
Politely decline (recommended)	The AI tells the visitor it doesn't have information about that topic. It won't guess or make things up.
Transfer to a human agent	The AI writes a natural, empathetic response, then automatically escalates the conversation to a human agent.
Answer anyway	The AI answers using its general knowledge, without your knowledge base context.

The Answer anyway option may produce inaccurate or made-up information about your products and services. Only use this if you're comfortable with responses that aren't grounded in your knowledge base.

In all cases, the AI generates a natural response in the visitor's language rather than using a static template message.

Agent Handoff

Control when and how visitors get connected to your human support team.

Automatic Agent Handoff: When enabled, the AI automatically transfers the conversation when a visitor asks to speak with a human agent. This only applies to visitor-initiated requests — it does not affect what happens when the AI doesn't know the answer (that's configured separately under the AI Identity tab).
Handoff Message: The message shown to the visitor right before they get connected to a human agent.
Show 'Talk to Support' Button After: How many messages a visitor must send in the last 5 minutes before the "Talk to support team" button appears. This prevents the button from showing too early in the conversation. Default: 6.

If you disable automatic agent handoff, we recommend setting up a Custom Workflow to handle agent requests instead. Without it, the AI might tell visitors it's connecting them to an agent without actually doing so.

Advanced

Fine-tune AI performance, privacy, and model settings.

Smart Answer Matching

Uses an extra AI step to find the most relevant information from your knowledge base before answering. Produces higher quality responses.

When enabled (recommended): Better answer quality, but adds approximately 1 second to response time.
When disabled: Faster responses, but the AI may sometimes use less relevant information to answer.

Privacy Protection (PII Masking)

Automatically hides sensitive personal data (names, emails, phone numbers, etc.) before sending messages to the AI. Keeps your visitors' private information safe.

AI Models

Shows the AI models powering this chatbot. These are managed by Yaplet and cannot be changed.

Primary Model: The main AI model used for standard responses.
Fast Model: A lighter model used for specific scenarios where speed is prioritized.

Danger Zone

Deleting a chatbot permanently removes all its settings and data. This action cannot be undone.

Use the Delete Chatbot button to permanently remove this chatbot. You'll be asked to confirm before the deletion proceeds.