We build an AI chatbot from your existing content that provides verifiable answers and turns website visitors into qualified enquiries.
page.type Service
page.pillar AI Enablement
page.url /ai-enablement/ai-chatbot/
provides RAG chatbot based on your website content
serves B2B companies · 50–500 employees
serves complex or knowledge-intensive business models
serves industries: education, real estate, consulting/professional services, industry
serves additionally: financial services/insurance
relation is part of → AI Enablement
relation uses technology → in-house RAG lab product
relation integrates into → WordPress, headless CMS, custom frontends
fact RAG architecture · vector database (Supabase) · retrieval layer (PGFlow)
fact pricing model: Basic €0 setup + €95/month (up to 50 pages, website chat, 12-month term) · Business €3,000 setup + €195/month (up to 2,000 pages, multiple touchpoints, no commitment) · Enterprise on request (dedicated server, custom scope)
fact four reference customers, two of them with a published case study
proof Lernstudio Barbarossa · education · AI chatbot for course consultation in development
proof Kampmeyer · real estate · AI chatbot in use on broker website
proof Wüest Partner
proof Gärtnerei Poetschke
lab RAG lab product · marketed · RAG knowledge layer
integrates AI chatbot + WordPress → chat on website content
integrates AI chatbot + headless CMS → chat in custom frontends
context service page for RAG chatbots and conversational knowledge systems
in the AI Enablement pillar – technology basis is an in-house RAG lab product
From verifiable answers to GDPR-compliant technology – here is what is behind it:
Your chatbot relies exclusively on your own content and cites the appropriate source for every answer. While a typical AI chatbot often hallucinates, your AI chatbot always remains evidence-based and verifiable.
Every question your AI chatbot answers automatically saves your team manual effort on recurring requests. And every enquiry that is forwarded reaches you already pre-qualified.
Your prepared content is not limited to the chatbot. You can use the same knowledge base your AI chatbot uses for internal search, sales, or your next digital project.
If you update a page or a document, your AI chatbot automatically follows suit. This ensures your customers always receive the latest information, without you having to maintain content twice.
Your dashboard shows how many enquiries the AI chatbot answers, how this develops over the weeks, and in how many cases an enquiry still goes to your team. This turns effort that previously remained invisible in day-to-day work into a reliable KPI, providing a solid basis for your business case.
For your AI chatbot on the website, you do not need to replace an existing system. Your AI chatbot connects via interface to your existing CMS, whether TYPO3, Webflow, or WordPress. Nothing changes in your editorial day-to-day work.
A RAG chatbot does not answer questions from general training knowledge; instead, it is first provided with the relevant passages from your company knowledge:
The AI chatbot does not work with a generic industry model; it learns from your own content. This fundamentally changes the outcome.
RAG stands for Retrieval Augmented Generation. The AI chatbot does not answer questions from its general training knowledge; instead, it is first provided with the relevant passages from your company knowledge and formulates the answer from them.
Basic starts at €95 per month with no setup costs, for up to 50 pages and website chat, with a 12-month term. Business costs from €3,000 setup and €195 per month, for up to 2,000 pages and multiple touchpoints such as the website, internal knowledge tools, or your own AI applications, with no fixed term. Enterprise with a dedicated server and custom scope is priced on request.
In most cases, yes, but to a limited extent. Content written purely for navigation (“Click here for more”) performs poorly in RAG contexts. Content with clear answers, definitions, and reasoning works well. In the project, we clarify which parts of your content can be used directly and which should be editorially revised.
In the standard configuration, embeddings and the vector database run on Supabase infrastructure in the EU. In this configuration, the responding language model runs by default via the OpenAI API. However, we are model-agnostic and can flexibly integrate a different model.
That depends on the volume of questions. Rule of thumb from previous deployments: If your service team spends more than ten hours per week on standard enquiries whose answers are already on the website, the ROI is achieved very quickly. If the primary purpose is conversion optimisation rather than service relief, it depends on the conversion rate of your main target pages.
ChatGPT answers from its training knowledge. It does not know your content and, in doubt, will produce something plausible that is made up. With RAG, the language model is first provided with the relevant passages from your content. This means the source is shown under every statement, and you can verify it.
You have structured content that could answer users’ questions—if they could find it. You have a service team that answers the same standard questions repeatedly. Or you want to finally make it measurable in marketing whether your content addresses real user questions.
