AI chatbots for your website: answers from your company knowledge

Chatbots answer visitor questions on the company's own website using the knowledge of the workspace assigned to them.

The chatbot is therefore the outward-facing side of the same knowledge base that is used in-house by Nex and the Document search as well. Whoever decides what a bot may see does so in the same place as for employees, namely in the Workspace. There is no second permissions system just for bots.

Replaces chat tools such as Intercom or Tidio with a bot that answers from your own company knowledge.

A chat window on a website is quickly installed. The work starts afterwards, because someone has to teach the tool what it is allowed to say. This is usually done with specially written text snippets that have to be maintained a second time alongside the maintained company knowledge. As soon as something changes, it changes in one place and not in the other.

The second breaking point is the question nobody anticipated. A tool that answers from prepared snippets cannot answer it and sends the person asking to the contact form. Yet the answer is often already available in-house, just in a document rather than in a chat snippet.

This section explains the core of the service, which is why it goes into detail. The difference lies not in the chat window but in where the text that appears in it comes from.

1. Step 1

   Create a bot and assign it to a workspace

   A new bot is given a name, a template, a greeting, quick replies and its system prompt. Assigning it to a workspace also determines which knowledge it sees and which it does not.

2. Step 2

   Embed on the website

   It is embedded via a script tag. The bot then appears as a widget on the page, with the configured name, your own colours and a pre-chat form, if one is wanted.

3. Step 3

   Answer questions, with source

   In the conversation, the bot answers visitor questions from the knowledge base of its workspace and provides the source reference along with the answer. Via connectors set up on request, it also draws on data from connected systems, such as availability or stock levels.

4. Step 4

   Book an appointment or hand over

   Directly from the conversation, the bot books an appointment and sends the confirmation email automatically. If it gets stuck, it hands over to a member of staff. Conversations and metrics for all bots are kept in the shared dashboard.

The component is not limited to support. The same mechanism powers a knowledge agent for specialist questions, appointment scheduling and an internal bot that provides information at a workstation or on the phone in field service.

• Support and frequently asked questions, answered from your own documents instead of maintained text snippets.
• Knowledge agent for specialist questions, internally and externally, for example in specialist departments and educational institutions.
• Appointment scheduling, from the conversation through the check to the confirmation email, in a single process.
• Information from connected systems, where an answer is held not in a document but in a database.
• Internal use at workstations and on mobile devices, when the question arises where nobody is sitting at a desk.

For the spoken channel, there is a separate component, the VoiceBot. It answers phone calls, while the chatbot stays with written conversation on the website.

• A bot answers from the knowledge base of its workspace. The more complete the stored content, the more questions it answers.
• How the workspaces are structured determines what a bot may see. This decision is made before the first conversation, not afterwards.
• It is embedded via a script tag on your own website.
• Appointment booking and confirmation emails require a stored email configuration for each bot, including the confirmation template.
• We set up connectors to databases, calendars, CRM, e-commerce, ERP and your own API on request. The system in question must be reachable and offer an interface.
• Operating a chatbot is subject to the transparency obligations of the EU AI Act; see the section on processing.

• One dashboard for all chatbots, with conversations and metrics in one place.
• Answers from the knowledge base of the assigned workspace, with source reference.
• Appointment booking in the conversation, with an automatic confirmation email based on your own template.
• Handover to a member of staff.
• Connectors to databases, calendars, CRM, e-commerce, ERP and your own API, set up on request.
• Freely configurable: name, colours, greeting, quick replies, system prompt and a pre-chat form.
• Can be embedded in any website via a script tag, as a widget alongside your existing site.