AI chatbot on your own data
Companies with many recurring questions from customers or staff whose answers are already in existing documents.
Updated: 2026-11
An AI chatbot on your own data answers questions from your documents, policies, product information or knowledge articles and cites the source. For customers on the website or staff on the intranet, with a permission concept, hosting in Switzerland or the EU, continuous quality measurement and the transparency notice the AI Act requires.
General-purpose chatbots invent answers when they do not know. For a company that is a risk: wrong promises to customers, wrong information to staff. At the same time, manuals, FAQs and policies hold the knowledge people search for every day.
A chatbot that answers only from these sources and cites them solves both, provided preparation, permissions and measurement are right.
What you get
Prepared knowledge base
Documents cleaned, structured and prepared for retrieval, with rules for updates.
Chatbot with citations
Answers only from retrieved passages, with a source reference and an honest notice when nothing is found.
Permissions and privacy
Access only to approved content, conversation storage under defined rules, model choice with a suitable region.
Quality measurement
Test set of real questions, regular evaluation and an overview of unanswered questions.
How it runs
- 01
Sources
Select content, find gaps and contradictions, define audiences.
- 02
Prototype
Build the chatbot with part of the content and measure it against real questions.
- 03
Expansion
Add all content, finish permissions, interface and integration.
- 04
Operation
Monitor quality, update content, use unanswered questions as signals for new content.
Price framing
Fixed price for prototype and expansion according to content volume; running costs for models and hosting transparent by usage.
Indicative, not a binding quote — the frame is confirmed during scoping.
Parent service: AI consulting & integration
FAQ
Doesn’t the chatbot make things up?
It is built to answer only from retrieved passages. Errors can never be ruled out completely, which is why we measure quality continuously.
Where do the models run?
Depending on requirements, with providers offering European regions or on infrastructure in Switzerland. Inputs are not used for training.
What applies under the AI Act?
Users must be able to recognise that they are talking to AI. An information chatbot is normally not a high-risk system.
AI chatbot on your own data
Companies with many recurring questions from customers or staff whose answers are already in existing documents.