Your documents and knowledge, made searchable

Answers from your own documents, with the source shown. Not from the internet, and not invented.

In short

 

In every company the knowledge is scattered. Part of it sits in procedures on the server, part in supplier manuals, part in old quotes, and a fair share in the heads of two or three people. The moment they are on holiday, the rest grinds to a halt.

An AI knowledge base solves part of that. Your colleague asks a question in plain language and gets an answer built from your own documents, with a note of where it came from.

Rights and roles stay intact

 

This is where such projects most often go wrong. If everything is dumped into one big pile, a colleague can suddenly read what is in a personnel file or an acquisition file through an innocent question.

So we connect the knowledge base to your existing access rights. Anyone who cannot open a folder gets nothing out of it through the AI either. That is set out in advance and tested before delivery.

What it is good at

Answering questions whose answer is somewhere in your documents, but nobody remembers where.

Getting new staff up to speed without a colleague being interrupted each time. More on AI for HR.

Supporting people on the road or on the shop floor.

Finding out what was agreed in older contracts and quotes.

HOW IT WORKS, IN PLAIN TERMS

  • Your documents are read in and split into chunks that hang together.
  • On a question, the system first looks for the chunks that are actually about it.
  • Only those chunks go to the language model, together with the question.
  • The model writes an answer based on what it was given, and points to the source.

That approach has a name, RAG, but what matters is what it means: the model does not invent from nothing, it summarises what is in your documents. If the answer is not there, the system should say so.

WHICH DOCUMENTS GO IN?

  • PDFs, Word documents, Excel files and presentations
  • Scanned documents, via text recognition
  • Content from SharePoint, a document management system or a shared drive
  • Manuals and technical sheets from your suppliers
  • Internal procedures, quality manuals and safety instructions
  • Data from your ERP, alongside the documents. More on that

WITH A SOURCE, ALWAYS

An answer without a source is worthless in a business context. Anyone deciding on the basis of a procedure has to be able to open that procedure. So every answer shows what it rests on: which document, which version and which passage.

GOOD TO KNOW

That is also your protection afterwards. In a dispute about what was agreed, you can show which document applied at that moment.

So we record which version of a document was used, and keep that.

30

YEARS OF .NET EXPERIENCE

1,000+

PROJECTS DELIVERED

100%

SUCCESSFUL ERP INTEGRATIONS

10

IN-HOUSE SPECIALISTS

What you need to start

01

A bounded set of documents. Not the whole server, but one domain you know is sound.


02

Someone who can say which version applies in case of doubt.


03

A decision on where those documents may be processed. More on AI, GDPR and the AI Act.


04

A handful of test questions whose correct answer you already know. That is how we measure whether it works.


What it is less good at

We would rather be straight about this. A knowledge base is weak when your documents contradict each other, because it then answers from the document that best fits the question, not from the right one. It is also weak on questions that require calculation or combining across many documents.

In practice such a project therefore almost always brings to light that outdated versions are still floating around. That is annoying, but it is a gain: you clear up what was already wrong.

 

 

Why companies choose IDcreation

  • The Belgian Defence has been our largest client for over ten years. What meets those security requirements will hold up with you.

  • No standard package you have to bend to, but software that follows the way you work.

  • Our own hosting in Belgium, with monitoring, backups and a single point of contact.

  • Since 1996 we have built in Microsoft technology. One strong foundation makes us fast and thorough.

Frequently Asked Questions

Do our documents stay ours?
Yes. We agree up front where they are stored and processed. Your documents are not used to train third-party models.
The risk exists but is strongly limited, because the system works only with the passages pulled from your documents, and because every answer shows a source you can check.
Technically a great many. In practice we advise starting small with one domain and expanding once the answers prove reliable.
Yes, through text recognition. Quality does depend on the scan. Poor scans give poor answers, and you notice that immediately.
Yes. Your colleague can ask a question in English about a German manual and get an English answer.
With a fixed set of test questions whose answers you know. We rerun those on every change, so you can see whether quality stays stable.

 

Is your knowledge scattered across too many documents?

Tell us which domain generates the most questions. We will look at what it takes to make it searchable.

 

Contact Us

 

 

 

This website uses cookies The IDcreation nv website uses cookies to keep our website reliable and secure, to measure performance and to offer personalized advertisements. By allowing cookies, you agree that we share information with third parties, such as our marketing partners, who may combine it with other information that you have provided to them or that they have collected based on your use of their services. If you refuse the cookies, we only use functional cookies. You can manage your preferences via our Cookie policy.
Show details Hide details
Allow selection Allow all cookies