IT provider: searchable knowledge, permissions intact
About this source
Gen8
Source: IT provider: searchable knowledge, permissions intact
Managed AI service for SMEs. We set up, connect and manage; you stay in control.
© 2026 Gen8 VOF · Dutch Chamber of Commerce 92978789 · Delft
Case study · IT services (management and projects for several clients)
The model must not see anything the requester may not see
Answer questions about working agreements, procedures and customer-specific setups from your own documentation, with sources and dates, without bypassing existing permissions.

Evidence status
Status: not yet approved for publication. Results on this page are only included with the customer's approval.
Starting point
Where it began
The pattern described by this case: organisational knowledge is scattered across document libraries, project folders and old notes, separated by client. Those who know the answer are asked; those who do not know it search or ask further. The information exists — finding it is the problem, and separation between clients rules out a simple search across everything.
Approach
What we built
This type of workflow answers a question using documents the requester may access, showing where each answer comes from: the document, the location within it and when it was last changed. This reference is central — it reveals whether an answer rests on outdated material.
If no adequate source is found, the answer is that there is no source. Nothing is supplemented from general knowledge, because a plausible answer about a customer-specific setup is worse than no answer.
Control
Where people stayed involved
The workflow is read only: nothing is edited, moved or added. Existing permissions remain decisive and are applied again to every question; a document the requester cannot open does not exist in their answer either.
Every question and consulted source is logged. This allows accountability for who requested what, and reveals consistently missing documentation — questions with no adequate source are exactly the topics not yet written down.
The workflows are under querying an internal knowledge base and making documents accessible as knowledge. Other cases follow the same structure.