Skip to main content

AI chatbot integration for your website

A website chatbot is only useful if it answers from your actual content, admits when it does not know, and hands over to a person at the right moment. We plan, integrate and review chatbots that meet these standards and the legal requirements that come with them.

ISO 9001 & ISO/IEC 27001Certified quality and information securitySince 2014, made in GermanyDiese Seite auf Deutsch

More than a widget in the corner

Dropping a generic chat widget onto a site takes minutes. Making it answer correctly about your products, opening hours, shipping terms or service conditions takes deliberate work: selecting sources, keeping them current, defining what the bot must not say and deciding where a human takes over. Without that, visitors get confident but wrong answers, which costs more trust than having no chatbot at all.

We build website chatbots on kalAIDoCreate, the enterprise AI platform from jocapps with governance, logging and custom APIs, or integrate them into your existing stack. Before launch, UXreviewer.io tests the conversation design and the widget itself with a structured review, because a chatbot is part of your user experience, not a separate technical project.

What to watch out for

  • Under the EU AI Act, visitors must be informed that they are talking to an AI system. Put this in the widget itself, not only in the privacy policy.
  • If the chat widget sets cookies or loads third-party scripts before consent, it falls under your consent management. Load it only after the relevant consent or use a setup without non-essential storage.
  • Visitors type personal data into chat fields, including order numbers, health details or complaints. Define retention periods for transcripts and make sure your processing agreements cover the model provider.
  • Prompt injection is a real risk once the bot can follow links or call tools. Restrict what the bot can do and never let it reveal internal instructions or data from other sessions.
  • The widget must be usable by keyboard and screen readers. Accessibility requirements for consumer-facing services apply to the chat as much as to the rest of the site.

What a well-integrated chatbot delivers

Answers grounded in your content

The bot retrieves from your pages, FAQs and product data instead of relying on general model knowledge. Answers can cite the page they come from so visitors can check.

Clear limits and handover

Topics like legal advice, complaints or pricing exceptions are routed to your team via email, ticket or live chat. The conversation history travels with the handover.

Governance for operators

With kalAIDoCreate you control which model is used, review conversation logs and adjust instructions without a developer. Access is role-based for marketing, support and IT.

Tested user experience

A UXreviewer.io review checks tone, error handling, mobile layout and accessibility. You receive concrete fixes before visitors see the bot.

From idea to live chatbot

  1. Use cases and sources

    We define the top questions the bot should handle and the content it may use. Out-of-scope topics are listed explicitly.

  2. Build and connect

    Sources are indexed, the widget is embedded and handover to your support tools is wired up.

  3. Review and test

    Test conversations, adversarial prompts and a UX review reveal gaps before launch.

  4. Launch and improve

    Unanswered questions and handover reasons are reviewed regularly to improve sources and instructions.

Matching products

kalAIDoCreate

Governed AI platform for enterprises and the public sector. An all-in-one AI platform for enterprises and public institutions with compliance features, governance and custom APIs.

Explore kalAIDoCreate

UXreviewer.io

Expert UX reviews with concrete action steps. Structured, expert usability reviews of apps, websites and chatbots with concrete action steps that improve retention and conversion.

Order a UX review

Frequently asked questions

Which content can the chatbot use?

Typically web pages, help articles, product catalogues and PDFs you approve. We exclude internal documents unless there is a clear reason and an access concept for them.

Can the chatbot run on EU infrastructure?

Yes, kalAIDoCreate supports model and hosting choices suited to EU data protection requirements. The exact setup depends on which models you want to use.

How do we stop the bot from making things up?

We restrict answers to retrieved sources, instruct the bot to say when it does not know and test with questions designed to provoke wrong answers. Logs show where it still struggles.

Can it also run inside our mobile app?

Yes. The same chatbot can be embedded in a web view or native screen of your app, so web and app visitors get consistent answers.

Let's talk about your project.

15 minutes with an AI Code Engineer, free and without obligation. Or call our team around the clock.