EDIH.li
DE
Back to knowledge

Case study · Enterprise knowledge

Private Knowledge AI Hub: secure access to internal knowledge

The CANCOM Assistant connects generative AI with approved knowledge sources. EDIH.li provides it as a Private Knowledge AI Hub. The KI@RA example shows its use at Raiffeisenlandesbank Upper Austria.

CANCOM Assistant · KI@RA · Raiffeisenlandesbank Upper Austria

CANCOM reference case KI@RA at Raiffeisenlandesbank Upper Austria

Browse the case study

1 / 4

Case study · Enterprise knowledge

Knowledge exists, but access takes time

Complex knowledge and process structures slow down daily work. Employees spend substantial time searching for information and interpreting it. The organisation also had limited practical experience with generative AI.

Knowledge exists, but access takes time

Complex knowledge and process structures slow down daily work. Employees spend substantial time searching for information and interpreting it. The organisation also had limited practical experience with generative AI.

How KI@RA supports employees

  • A digital colleague makes approved internal knowledge accessible through a chat interface.
  • The assistant accelerates knowledge transfer across the group.
  • The organisation gains practical experience with generative AI and large language models.
  • The findings contribute to a medium- and long-term data and AI roadmap.
  • Wikis, SharePoint, Confluence, documents and individual sources can be connected.

Private AI with architectural choice

Companies can choose cloud, their own Azure environment, on-premises or a hybrid architecture. They also select the language model. SSO, Microsoft Teams integration and feedback support daily use. CANCOM states that the solution is hosted in the EU and does not charge licence fees per user account.

Measure value during the pilot

The CANCOM material provides no measured time or cost savings for this reference case. A pilot should therefore track search time, answer quality, usage and follow-up questions. Source: CANCOM, ‘CANCOM Assistant Introduction’, 16 September 2026.

Would you like to implement a similar use case? We will help define the right first step.