HUS Kiinteistöt decided to organize a competitive tendering process regarding their purchase invoice system, when their then current license software was coming to the end of its lifecycle. At the same time, the potential of AI in purchase invoice handling was recognized, and the use of AI was included in the competitive tendering requirements.
HUS Kiinteistöt selected the CGI Purchase to Pay module in the CGI DataCycle360 portfolio for their purchase invoice handling. The handling process in the Purchase to Pay solution enables the processing of all purchase invoice types at the same time with the help of AI.
“We noticed that with AI we can make the processing of purchase invoices more efficient, reduce the amount of expensive manual work, and direct the work of our experts towards development. This has been a great advantage in our organization. Nowadays, only a few invoices are handled manually”, says Alexander Kozlov, Leading Specialist of Digitalization in HUS Kiinteistöt.
“The project was carried out perfectly. It is a textbook example of how an IT project can be run on time and on budget.” - Alexander Kozlov, Leading Specialist of Digitalization, HUS Kiinteistöt Oy

Tireless AI brings continuity and reliability
About 80% of purchase invoices in HUS Kiinteistöt are routed with AI, leaving 20% of the work to be carried out manually or by robotic process automation. For example, with new vendors robotic process automation will identify the person responsible for the subject check on the invoice and then route the invoice accordingly. Altogether 76% of one person’s working time is released from purchase invoice routing to other tasks.

A purchase invoice to be processed by AI needs to contain all mandatory details. When it has been ensured that these mandatory details exist on the invoice, AI will process the invoice based on history data and calculate probabilities to the subject checker, approver and for posting dimensions.
“AI can calculate probabilities and suggest postings, resulting in a very high success rate in purchase invoice handling. The manual work done now is mostly related to checking the invoices AI has processed”, states Alexander Kozlov.
The routing costs of invoices processed by AI are smaller than those caused by manual work. In addition, AI is never sick, on vacation or tired, and it is available 99% of the time, thus guaranteeing continuity and reliability.
From successful roll-out to training AI
In the project with HUS Kiinteistöt, the system was deployed right on time and on budget. The client’s expectations were met when all required features were included in the system, and the functionality was further complemented with a few additional features delivered later.
“The project was carried out perfectly. It is a textbook example of how an IT project can be run on time and on budget. CGI’s excellent project management practices ensured that the project went ahead smoothly, and integrations were implemented precisely according to our needs. All our requirements were assessed and included in the new solution”, applauds Alexander Kozlov.
When talking about the future, Alexander Kozlov says: “We are planning to ask our vendors to provide us with more detailed invoice messages, so that we can train AI to process purchase invoices even more efficiently. AI develops constantly, which will improve the accuracy of purchase invoice handling. AI learns and adopts our ways of working by imitating the work of actual persons.”
