AI has made investigating mistaken payments 90 per cent more efficient at Hrvatski Telekom, CFO Matija Kovačević said. The example featured in a Finance Weekend discussion in Rovinj, Croatia, on whether AI will reduce finance teams or change their work. Executives described automation alongside continuing needs for specialist judgement.
What does AI do with mistaken payments?
Kovačević said Hrvatski Telekom is deploying AI across finance and that this is changing the company’s work. He cited “ghost payments”, which are payments registered with a mistake, such as a wrong reference number. People previously investigated those cases, whereas AI has made the task 90 per cent more efficient, he said. The account concerns one type of payment check and does not give a broader measure of finance automation.
Why do executives expect finance roles to change?
Kovačević said AI would change finance teams rather than replace them. He identified data collection as the area likely to see the biggest change in finance departments. In his description, the value of the work is shifting from production towards interpretation. That distinction means teams may spend more effort scrutinising information produced by systems and less on gathering it manually.
The panel’s discussion also addressed AI’s impact on Hrvatski P&L. Kovačević said experimentation was currently increasing costs, while better deployment could make AI revenue-generating over time. He did not provide a value, timetable or forecast for that potential revenue. These remarks describe a current cost and a future expectation, rather than a reported financial return.
Could finance teams become smaller?
Kovačević said finance teams would be smaller in 2030 and take a more analytical role, examining AI data. He also said the value of finance departments would not decline. His view connects a possible reduction in team size with a change in the type of work remaining. The discussion did not specify which roles or tasks might be affected.
Bogdan Jelić, an account manager at Comtrade System Integration, said he did not think AI would shrink AI teams. He argued that each department has its own specialism, including finance, which AI cannot replace. He nevertheless said AI can automate some finance department processes. His comments underline that automation of tasks and replacement of an entire specialist function are different propositions.
What skills did Farseer’s co-founder emphasise?
Luka Mijatović, co-founder and CFO of Farseer, warned that AI can encourage finance teams to become reliant on easy prompts. He urged current and future finance professionals to learn the details of finance and show “willpower”. He said reliance on AI could compromise career progression. His contribution focused on how people develop expertise while tools handle a growing share of routine work.
Mijatović also described AI as a high cost at present, while saying it was increasing production substantially for engineering teams. No amount for the cost or measure of that production increase was reported. The comments leave open how much the effect varies between finance and engineering work. The source account gives no additional figures to compare those areas.
What remains uncertain after the discussion?
The panel took place at Finance Weekend, where Croatian founders and finance executives also discussed engineering talent and the challenges of launching startups in Croatia. The available account does not give detailed conclusions on those other subjects. Nor does it provide a forecast for staffing changes or a financial estimate for AI’s longer-term contribution. Kovačević’s remarks about 2030 should therefore be read as his view of how finance work may develop.
The clearest reported measure is the 90 per cent efficiency improvement for investigating mistaken payments at Hrvatski Telekom. It applies to that described task, not to finance work as a whole. Further deployment and reported results would help show whether the experimental costs give way to revenue, as Kovačević expects. No such outcome or timetable was provided in the discussion.