HomeAI chatbots
AI chatbot development A chatbot that knows your company and hands over when needed.
A good chatbot for business answers questions from your own documents, cites the source and knows when to hand over to a person. As an AI chatbot agency we build it for your website, WhatsApp and your team, on servers in the EU.
Free, no strings attached. If you like, with a written assessment of your project.
Four chatbots we often develop.
-
Chatbot for your website
Answers questions about products, services, delivery and returns from your documents and guides visitors to the right page or to contact. More under AI customer service.
-
Chatbot for WhatsApp
Answers enquiries where your customers already write, through the official WhatsApp Business interface, with handover to your team.
-
Knowledge chatbot for your team
Answers employees' questions about standards, manuals, policies, IT and HR, with sources, and pays attention to who may see which documents.
-
Chatbot for sales
Takes enquiries, asks the necessary follow-up questions and hands your sales team a complete, pre-sorted enquiry.
If the chatbot should not only answer but also complete tasks in your systems, we speak of an AI agent.
What makes a chatbot reliable in daily use.
-
It stays on topic
We define what the chatbot gives information about and what it does not. Questions outside that scope it either does not answer or hands over.
-
It stays up to date
When prices, opening hours or policies change, the knowledge base is updated. We set it up so that this runs in your company without extra effort.
-
It sounds like you
We agree tone, form of address and handling of complaints with you so that the chatbot fits your brand.
-
It hands over cleanly
When it is unsure, or when the user asks, the conversation goes to a person with a summary.
-
It shows what is missing
The evaluation shows which questions are asked and where there is no good answer. That also improves your knowledge base.
-
It is recognisable as AI and accessible
The chatbot identifies itself as AI (Art. 50 AI Act). We build the chat window so that it can also be used with a keyboard and a screen reader. More under accessibility.
How we develop an AI chatbot
-
Step 1: Your message
You tell us what the chatbot is meant for and who will use it. We get in touch with you personally.
What you do: write to us. There is nothing you need to prepare.
-
Step 2: Clarifying knowledge and goals
We look at your documents and typical questions and define what the chatbot answers and when it hands over. After that you receive a written proposal.
What you do: show documents and example questions.
-
Step 3: Build and test
We build the chatbot, connect it to your website, WhatsApp or your systems and test it with real questions.
What you do: test and give feedback.
-
Step 4: Launch and operation
The chatbot goes live, at first with close evaluation. We adjust, extend the knowledge and add further channels on request.
What you do: give feedback.
Your data stays your data.
- Stored in the EU. Wherever possible, the knowledge base and conversation data are held on servers in the EU. Our servers run at AWS in Frankfurt. Where that is not possible, we tell you beforehand.
- Under contract. You receive a data processing agreement, as the GDPR requires. We define retention period and deletion with you.
- No training on your data. Your documents and conversations are not used to train language models.
Frequently asked questions about chatbot development
How does the chatbot answer from our documents?
With a technique called retrieval-augmented generation, RAG for short. For every question, the chatbot finds the relevant passages in your documents, writes the answer from them and cites the source. If nothing fits, it says so and hands over. Which documents it may see depends on who is asking. Your knowledge stays in your own data storage and can be changed or deleted at any time.
What is the difference between an AI chatbot and an AI agent?
A chatbot answers within a conversation. An AI agent can also complete tasks in your systems, such as creating an order. Many projects begin with a chatbot and grow into an agent.
Why not simply embed ChatGPT on the website?
A general language model does not know your company and answers even when it knows nothing. A chatbot built on the RAG principle answers from your documents, names the source and hands over when it is unsure. On top of that come questions about data protection and contracts.
How do you prevent the chatbot from inventing things?
It answers only from the passages found in your documents and names them. If there is no reliable passage, it says so and hands over. We test it on your real questions before launch and keep measuring in operation. Nobody can guarantee freedom from errors, but errors can be limited and found.
How does the chatbot's knowledge stay up to date?
We set up the connection to your sources so that changes are picked up, for example from your website, your document storage or your knowledge base. What you would otherwise have to maintain, we clarify beforehand.
In which languages does the chatbot answer?
In the language of the question, even if the documents exist in another. Which languages you offer we decide together, and we test them before launch.
How is the chatbot embedded in our website?
As a chat window that matches your website in design and language. We take care of installation, testing and the connection to your systems.
Where is the data stored?
Wherever possible on servers in the EU, with a data processing agreement. More under AI and data protection.
What does it cost to develop an AI chatbot?
That depends on the volume of documents, the channels and the connected systems. A price list would therefore tell you little. Before we begin, you receive a written proposal with a clearly described scope. Without your approval, no costs arise.
We only want a chatbot. Are we in the right place?
Yes. An AI chatbot for your website, WhatsApp or email can be set up on its own and is often the first step. It answers from your documents, identifies itself as an AI and hands anything unclear to your team. If it should later also look up orders or create tickets, it grows into an agent for customer service, without starting over.
What should your chatbot give information about?
Send us a few sentences. We will reply personally, free of charge and without obligation.
Does ChatGPT already name your business? Check it for free before we talk about your chatbot.