An AI chatbot for your website and customer enquiries
A website chatbot gives visitors a place to ask questions while they explore your business. With WeBot, prepare an agent using approved knowledge and instructions, then review the available website widget or custom chat integration. Start with a useful information task and test the answers before making the assistant part of your customer journey.
Identify the questions visitors need answered
Use actual customer enquiries to choose the first task. A visitor may want to understand a service, check whether it fits their needs or find the next step for contacting the business. Group related questions and write down which source answers each one. Keep account-specific or confidential requests out of a general information flow unless an appropriate authenticated integration is configured. A focused assistant is easier to test because you can compare its responses with a defined set of approved facts.
Prepare knowledge that stays useful
Review the material you give the assistant before publishing it. Remove outdated offers, contradictory instructions and details that should not be shared with a website visitor. Explain business terms in ordinary language and include the conditions that change an answer. If the chatbot cannot verify a detail, its instructions should tell it to say so and direct the visitor to the appropriate next step. Update the source whenever the underlying service changes.
Choose a widget or a custom chat interface
WeBot provides a ready-made website widget and a headless JavaScript integration for a custom chat interface. The widget can present a floating or inline chat panel, while a custom interface lets your application control the surrounding experience. Review the signed-in Chatbot SDK and API documentation to choose the supported approach. Custom chat requires a backend that creates owned sessions securely. Keep private API credentials on that backend rather than placing them in public page code.
Make the first message clear
Tell visitors what the assistant can help with and make its purpose easy to understand. Avoid asking for unnecessary personal information before someone has explained their question. Use a short opening, then test how the assistant handles follow-up questions and corrections. When it cannot answer, it should offer the appropriate business contact or next step defined in your instructions. The first-message and system-prompt guides below help you prepare this behaviour before configuring the agent.
Use personalisation only with appropriate context
Dynamic variables can supply relevant context to a conversation when your integration supports them. Plan the variable names and values your backend will provide, and test what happens when a value is absent. A name can make an introduction clearer; it does not establish permission to reveal account information. Keep ownership and session authentication in your application flow, and limit the context to what the intended task requires. Review the public dynamic-variable guide before building your integration.
Test the complete visitor experience
Try the chatbot on a phone and a desktop. Check whether the chat control is easy to find, whether long answers remain readable and whether visitors understand connection feedback. Ask short questions, incomplete questions and questions outside the approved knowledge. Verify that a repeated click or a refreshed page behaves as your chosen integration documents. Inspect the saved conversation evidence available in your owned account, then fix the knowledge or instructions that caused an unclear answer before expanding the rollout.
Review the service before publishing
Compare the current plans and usage settings, then sign in to configure the agent and integration. Technical integration references and API-key management require login, while this overview and the setup guides are public. Begin with a controlled deployment and review feedback from the intended audience. A chatbot can explain approved information and guide a visitor, but your business should still provide an appropriate route to a person when the request needs individual attention.
Explore agent services and setup guides
- Voice agents for business conversations
- Website chatbot and chat integrations
- Write an agent system prompt
- First messages and AI disclosure
- Personalise conversations with dynamic variables
- Prepare a batch-call CSV file
- AI Agent Terms of Service
Agent setup, saved conversations and integration credentials require sign-in. Public guides explain the service before you create an agent.
Common questions
Can I add a WeBot chatbot to my website?
WeBot supports a ready-made website widget and a custom chat integration. Sign in to review the current SDK documentation, configure the agent and connect your site.
Do I need a backend for custom chat?
The headless integration uses your backend to create owned sessions securely. Keep API keys on the server and follow the current signed-in integration reference.
Can I provide my own business knowledge?
Prepare approved business source material in the agent workspace and test the questions it should answer. Keep the information current and define how to respond when an answer is unavailable.
Where do I start?
Read the public system-prompt, first-message and dynamic-variable guides, compare plans, then sign in to configure and test your agent.