AI chatbot development
What I Mean by an AI Chatbot
An AI chatbot answers questions from your visitors or staff in everyday language, directly on your website or in your web application. I build it for small and medium-sized businesses in the DACH region that already run a website, a shop or a web application and want to hand a clearly defined task to an AI feature.
I rely on existing AI services, meaning API and cloud providers, including ones from the EU. Which service suits your project we decide together, see AI and Hosting in the EU.
Chatbot or Knowledge Bot
The two terms are often mixed up. I separate them like this:
- Chatbot: holds a conversation on a fixed topic, for example opening hours, how an enquiry works or guiding questions for choosing a product.
- Knowledge bot: answers questions from your own material, such as manuals, FAQ pages, product data or guidelines.
How a Knowledge Bot Works with RAG
RAG stands for retrieval-augmented generation. The language model does not answer from memory; it first receives matching passages from your documents. The process:
- Prepare: Your content is converted into vectors, number sequences that represent the content.
- Store: The vectors sit in a vector database that can be searched by closeness of meaning.
- Search: When a question comes in, the system fetches the best-matching sections.
- Answer: The language model writes the answer based on those sections. On request the bot shows the source.
The advantage: new or changed documents can be added without retraining the language model.
What I Build
- Interface: A chat window in your design, accessible and usable on mobile devices.
- Connection: AI service, vector search and your data sources, with clear error handling when an answer is missing or unusable.
- Limits: We define the topic area, tone and behaviour for questions outside the topic.
- Handover: When unsure, the bot points to a contact form or a contact person.
- Tests: I secure core functions with automated tests.
Realistic Expectations
I do not want to raise false expectations: a chatbot can be wrong, even with RAG. Searching your documents makes answers easier to verify because the source can be shown. It does not give complete certainty. For information with legal or financial effect, I recommend using the bot only as a pointer and leaving the decision with your team.
Maintenance and Operations
On request I take care of hosting, updates and further development, see Hosting & Maintenance. This includes adding new documents and adapting the bot when your offering changes.
How This Differs from Related Services
- If an AI should read documents and let a person review the result in an editor, AI document processing fits.
- If an AI should carry out tasks in your systems on its own, AI agents and MCP servers fit.
Discuss a Chatbot
Tell me briefly which questions the bot should answer and where the answers come from. Then we check together whether a chatbot or a knowledge bot fits and how it slots into your system.
Related Services
- AI Integration - all AI services at a glance
- AI for WordPress and TYPO3 - building AI into existing content management systems
- AI and Hosting in the EU - location, model choice and third-party services in view
Frequently Asked Questions
What is the difference between a chatbot and a knowledge bot?
A simple chatbot answers only from what the language model knows on its own. A knowledge bot first searches your documents and writes the answer based on what it finds. This method is called RAG (retrieval-augmented generation). For questions about your products, price lists or manuals, a knowledge bot is usually the better choice.
Can the bot give wrong answers?
Yes, a mistake remains possible. Searching your documents makes answers easier to verify but does not replace a check. That is why I build in limits: the bot only answers the topics you have stored, shows the source on request and refers to your team when it is unsure. I do not promise a percentage.
What does an AI chatbot cost?
That cannot be answered with a flat figure. It depends on the amount and condition of your documents, the connection to your system, the languages, the expected usage and whether I take over operations. After a short conversation I can estimate the effort and make you an offer.
Can the chatbot be built into my existing website?
Usually yes, whether WordPress, TYPO3, an online shop or a custom web application. I look at your system and choose the route that can be maintained with reasonable effort. For WordPress and TYPO3 I describe this in more detail under AI for WordPress and TYPO3.
What about data protection?
I take GDPR requirements into account in the technical implementation; whether a specific use is lawful is assessed by your data protection advisor. This is not legal advice. We decide on the location and the model together before I build anything.