From insurance payouts to banking, AI is taking over tasks that once took hours or days of human work – which office jobs are first in line for automation?
Two years ago, processing a health insurance claim at one of Greece's largest insurance companies took on average 14 days. Medical opinions, test results, invoices and policy details all had to be checked and cross-referenced before payment was given the green light.
ADVERTISEMENT
ADVERTISEMENT
Today, the processing time has been cut to around one minute and the compensation is paid out the following day.
Driving this change is AI Flows, an artificial intelligence tool for document processing and workflow automation developed by the Greek company Archeiothiki. The system handles around 1,500 claims a day, "reads" the documents, extracts the necessary information and cross-checks it against the policy and the insurance company's rules.
Above all, though, it enables something that was previously practically impossible: checking 100% of cases rather than just a sample. This makes it possible to spot potential fraud that would otherwise have gone unnoticed.
The example of the Greek insurer offers a snapshot of a much broader transformation unfolding in offices across Europe.
From 16 hours to a few minutes
In banking, similar technology is used to assess the ESG (environmental, social governance) criteria of companies seeking finance.
The process involves around 300 questions. To verify the answers, staff must hunt for data in company reports running to hundreds of pages, ranging from the ratio of men to women to information on environmental performance and corporate policies.
This check used to require about 16 hours of human labour. Now, an AI system can search for the necessary information even in reports running to 400 or 500 pages, identify the relevant passages and process them in a fraction of the time.
The final assessment of ESG criteria influences the interest rate at which the company will be financed.
Jobs on the frontline
Archeiothiki currently applies AI Flows to around 100 different workflows for ten clients, mainly banks, insurance companies and businesses in the healthcare sector.
But what is happening in Greece is far from unique.
Office roles that have traditionally formed the back office of companies are among those most exposed to the new generation of artificial intelligence.
According to the report "Generative AI and Jobs: A Refined Global Index of Occupational Exposure", published in May 2025 by the International Labour Organization (ILO) and the Polish research institute NASK, administrative and clerical occupations remain the category of jobs most exposed to generative AI.
Among the most exposed roles are data entry clerks, staff involved in accounting and bookkeeping, and administrative secretaries. In high-income economies, roughly one in three jobs shows some degree of exposure to generative AI.
Artificial intelligence has already entered European workplaces on a large scale. According to the study "Digital Monitoring, Algorithmic Management and the Platformisation of Work in Europe" by the Joint Research Centre of the European Commission, published in October 2025, 30% of workers in the EU already use AI tools in their jobs. Use is particularly high in office work.
The transition also has a clear gender dimension. More recent ILO research shows that occupations where women are in the majority are almost twice as likely to be exposed to generative AI as those where men predominate: 29% compared with 16%.
One key reason is the greater presence of women in administrative, clerical and support roles.
"I will not hide it. There are jobs that are lost too," was the response of Archeiothiki executives when asked about this by Euronews.
The picture they sketch, however, is more complex. In many cases, automation is not used simply to carry out the same volume of work with fewer people. It allows companies to take on tasks that were previously practically impossible to perform at this scale.
One client, for instance, had around 50 people working on a single process and still built up a year-long backlog. In other cases, where only a sample of cases could previously be checked, now the entire set can be examined.
This aligns broadly with the ILO's assessment. The organisation warns that the "exposure" of a job to artificial intelligence does not necessarily mean that role will disappear. Few occupations consist solely of tasks that can currently be fully automated. For this reason, the most likely outcome for now is the transformation of jobs rather than their complete replacement.
The next data reservoir: the public sector
In Greece, the next major field may lie in the millions of documents that have already been digitised in the public sector.
According to Archeiothiki's chief executive, Andreas Papadakis, the company has carried out around 15% of the relevant digitisation projects for the state.
The problem is that much of this new digital archive remains effectively "blind". Physical documents were turned into digital files and given basic metadata so they could be searched. Their content, however, was not necessarily converted into structured data that can be analysed and put to use.
The first stage of digitisation was to turn mountains of paper into digital files. The next stage is already under way.
AI systems are no longer confined to reading and recording documents, but can extract information, cross-check it and execute entire workflows.
This shifts automation away from repetitive data-entry tasks to a much larger portion of the traditional back office.
