Published September 30, 2026 in Artificial Intelligence • 8 min read • 
Audio available
For Raiffeisen Bank International (RBI), analysts need to assess the credit risk of clients to ensure profitability and comply with strict regulatory requirements. But integrating a company’s financial data, its development over time, reasons for changes, and other relevant information can take days for quarterly and annual reporting. In November 2022, as the launch of ChatGPT dominated news headlines, it struck Hannes Mösenbacher, Chief Risk Officer at the Austrian bank, that the advent of AI presented an opportunity.
He challenged Bettina Köppe, Head of Group Customer Risk Intelligence, and Gernot Hinterleitner, Head of Group Corporate Credit Analysis, to use artificial intelligence to build a way to make their data “talk” as they evaluated the credit risk.
Just two years later, RBI was recognized as a leader in AI implementation in finance, winning three awards in the 2025 Global Finance Awards, including the overall award for Western Europe with the Risk Assistant. Building on the foundation of this AI-supported Risk Assistant, RBI successfully extended the underlying approach to other areas and products, from automated document verification and fraud detection to know-your-customer requirements, among other areas.
The transformation started with a straightforward challenge: how can we improve the overall process, timelines, and outcome of risk assessment? The team told us how they developed the bank’s innovative Risk Assistant.
For RBI, integrating company financial data, its development over time, reasons for changes and recent updates, and other relevant information took days for reporting and continuous monitoring. Producing these reports required a team of 20 analysts to manually collect data from multiple systems, interpret regulatory requirements, and compile detailed summaries in order to produce approximately 8,000 rating reports for corporate clients each year.
The team believed that if they could develop a minimum viable product with the Corporate Rating Reporting Assistant, this idea and approach could then scale to other process steps, customer types, and areas in collaboration with Customer Risk Intelligence.
By applying standardized data processing and document analysis, it minimizes the risk of human error and ensures that reports follow regulatory and internal governance requirements.
As part of its broader digital transformation strategy, the bank has developed an AI-supported Risk Assistant designed to automate the preparation of internal risk reports. The team identified the most promising opportunity, which would both reduce the time required to write risk reports and improve their quality.
The AI Risk Assistant reduces the workload by automatically gathering relevant information, analyzing documents, and generating draft reports that once took more than a day’s work for an analyst in just three hours. The system leverages generative AI technologies hosted in a secure environment, ensuring that sensitive financial and regulatory data remains protected while still enabling advanced automation capabilities.
Beyond simple automation, the Risk Assistant also helps improve consistency and accuracy in reporting. By applying standardized data processing and document analysis, it minimizes the risk of human error and ensures that reports follow regulatory and internal governance requirements. This is particularly important in banking, where risk reporting must be detailed, timely, and compliant with strict supervisory rules.
By streamlining this complex internal process and reducing the time required to produce risk reports, the system allows analysts to focus more on interpreting risks and making strategic decisions rather than performing repetitive document production.
And at first, the quality was about the same as a human-generated report. However, as the team iterated on the solution, they began to include a greater volume of information in the reports. The team realized, too, that there was an impartiality to the machine’s inclusion of data: while humans may be biased for “good” news, the Risk Assistant collected all data, meaning that reports were more thorough and nuanced.
Transform your career and business with next-generation AI and digital skills
Having built a highly functioning machine, the team can now use the same base knowledge to expand across other areas. For example, after the US and Israel launched air strikes on Iran in late February 2026, the team was able to set up a prompt in one day that would analyze what impact this might have on its clients, quickly producing reports on first and second-round effects. What would have taken weeks in the past they could do in hours, predicting what impact jumps in gas prices and interest rates would have on their clients. Mösenbacher observes that the newer LLMs help with liquidity risk management; ChatGPT helped with language, but the newer models, like Claude, work with structured data, which is helpful for market and liquidity risk.
The question “What is the revenue of company X?” may sound straightforward, but it was not easy for early versions of LLMs. If a machine finds revenue and change, it can think this is what you want, but the company may have four large segments, so getting the AI to see the big picture was a challenge when the team started the project in late 2022. Just three years on from this early development stage, the team says, LLMs are much better at recognizing this. “But I’m not sorry we had this journey,” says Mösenbacher, “because it gave us time to work on culture and manage change at RBI.” If people help shape the new environment, their fear is mitigated somewhat because they can see, and help build, a better-performing solution.
And, indeed, if you look at engagement of corporate analysts throughout the project, you can see a consistently high level of participation throughout the PoC, MVP and adoption stages of the project. Having built a culture of inclusion throughout the project and having emphasized the notion of “doing more” rather than replacement, RBI has retained the same number of employees at the Corporate Analysis Area as it had at the beginning of the project. However, within the Corporate Analysis Area, some employees have added “expert prompt engineer” to their list of skills.
Producing a risk report should be a starting point; analysts should ingest the information so they can make decisions.
Addressing the challenge of increasing complexity and inflow of information is critical in risk analysis today: if a machine can process the information, the human analyst can focus their energy on using the information to make recommendations. This points to a fundamental question in the age of AI: “What are you hiring highly paid experts for?” Is it to perform repetitive tasks?
Producing a risk report should be a starting point; analysts should ingest the information so they can make decisions. “People should be hired not to do the paperwork, but to do the thinking,” observes Mösenbacher, explaining: “If 70–80% of time is spent on routine work, and only 20% is thinking, you are actually not getting enough from your people, and your people are bored. We need to keep the joy in work!”
With the additional capacity, analysts can focus more on other relevant tasks, including industry analysis, geopolitical impacts on the customer, or on environmental, social, and governance (ESG) issues and assessments of business models.
New LLM models calculate based on skills: the law is the guiding principle, but it can be translated into operational code. With normal human language, you add some formulas, give a few examples, and create a basic version of a “skill” that can be the basis for the development of a product. With the required infrastructure already in place to scale different AI use cases, experimenting and prototyping can be accomplished more quickly and easily, enabling a rapid group-wide rollout. “You can deploy in your environment, and then users have access in the blink of an eye; a person can have an idea in the morning, and the team can be using it by the afternoon. Two weeks later, it can be productive; we no longer need two years.”
Proper and prudent credit and liquidity risk management complying with the strict regulatory requirements is imperative, and it must reflect the dynamic market movement, delivering a fast and comprehensive scenario analysis. RBI now has up-to-date AI technology at hand, which is uniquely capable of covering these scenarios immediately, providing deep insights to the team.
Once primarily the guardian of traditional and well-established legal systems, today the CLO is becoming the architect of enterprise artificial intelligence (AI) governance, helping to identify and manage AI risks and opportunities...
Too many leaders ask how GenAI will impact their businesses and products. Instead, they should ask: How can AI change our customers’ motivations, abilities, and opportunities?...
Audio available
Question central planning: why a unit-focused hybrid approach may outperform both centralized control and chaotic AI adoption....
Can a CEO's leadership be delegated to an AI avatar? Explore benefits, risks, and guardrails shaping digital leadership....
Explore first person business intelligence from top minds curated for a global executive audience