Fintech – Pioneering use of generative AI in banking and finance

Share this post:

Urban Roth, IBM Innovation Studio Leader, Stockholm

The race for leaders to meaningfully leverage generative AI and its transformative power for their organizations continues.

Current state of the banking industry – approaches, perceived opportunities and risks

A recent IBM survey of 600 banking executives, “World Outlook for Financial Services in 2024,'' found that tactical approaches that explore unique use cases are currently the norm, with 8 out of 10 I support the approach. The business area identified as having the greatest potential is risk and compliance reporting, closely followed by customer engagement. However, more than 50% of executives are hesitant to implement risk and compliance solutions, likely due to concerns about governance and AI-related risks. Top concerns include cybersecurity vulnerabilities, legal liability, and accuracy and bias assessments. This hesitation and risk perception help explain why digital assistants are the most popular starting point.

How can we stimulate broader applications of AI while addressing the risks involved?

Industry pioneers show the way!

In an effort to do just that – inspire the industry by learning by doing – we decided to collaborate with industry pioneers in fintech. We designed IBM Fintechx, a program where IBM experts validate use case ideas and collaborate with fintech groups to create prototypes. We will then share our results with the industry at Demo Day. So, learn and inspire the industry while contributing to your own ventures and existing services. To select the most interesting ideas and disseminate positive results, we assembled a group of industry experts from banking, investment and industry organizations (Handelsbanken, Wellstreet, Ålandsbanken, Resurs, Crosskey, Findec). Seven fintech companies ultimately invited to join IBM Fintechx: Edger Finance, Esgaia, Flowpay, Oxide.AI, TrustAnchorGroup, Asteria, Finterai.

So what are the applications developed in collaboration with fintechs?

Applications from Fintechx collaborators

Five of the results of the Fintechx project have been published and are briefly summarized below. A complete report on aggregated learnings and identified design patterns is available at:

Generative AI Fintechx Applications:

1. edger finance Our mission is to create EU financial market information comparable to the US EDGAR by collecting and aggregating all stock market information. Their prototype focused on automating data extraction for a company's quarterly reports and summarizing key points from the 30-page report. Here the team had to work with the accuracy of the data in mind when summarizing. The solution relies on IBM watsonx.ai and watsonx Assistant. Read more, case study.
2. flow pay started its journey with a lot of data and an idea: to provide simple online operating capital to small and medium-sized enterprises (SMEs). To enter a new market, the cash flow-based risk scoring model needed an initial dataset to interpret and classify the country's PSD2 transactions. The prototype was developed using an extensive language model that automatically generates its dataset from publicly available data. Flowpay's underwriting process also covered a prototype that generated an overview of an applicant's financial health and associated risks. The summary was in a non-technical format and was based directly on her PSD2 transaction data from the applicant's bank account.Read more, case study
3. Oxide AI By creating social media like an investment feed for our users, we want to help all of us find exciting investment opportunities. The feed is created by an AI agent trained in quantitative and qualitative analysis that traverses and analyzes the stock market to identify less obvious opportunities. For the new solution prototype, he moved from his GPT-4 on watsonx.ai to the Llama 2 model. It is an important step towards the use of open models suitable for enterprise-grade production. Reduced average continuous response time by 37% while maintaining quality levels. Read more, case study.
4. trust anchor group We help financial institutions and businesses monitor or invest in real-world assets by providing asset digitization technology. They built a dynamic asset valuation service using watsonx.ai that can calculate asset values ​​in real time. This is done by querying unstructured and structured data sources. It also allows business users to ask questions about their assets and explore valuation results. They also worked to ensure the accuracy of the data summaries. Read more, case study.
5. asteria has the ambition to provide small and medium-sized businesses with access to powerful cash management and a high degree of resilience in their business operations. They partner with existing banks to provide their services. Asteria has created an online advisor that can summarize a small business's financial situation, answer follow-up questions, and recommend bank financial products relevant to the small business's current needs. This advisor has the potential to further strengthen the resilience of SMEs while saving banks time. Read more, case study.

So what were the common characteristics across the projects, and what can we learn?

Common patterns offer disruptive opportunities beyond customer engagement

We conclude from these projects that this technology presents disruptive opportunities within and beyond the realm of client engagement, demonstrating the potential seen by the industry executives surveyed. It comes with a repeatable design pattern that can be applied across the financial business field and poses risks regarding the accuracy of results that must be managed with competence and care.
A wide range of applications are all derived from the following gen AI capabilities:
– Obtain useful information from unstructured data
– Transform structured data and information into knowledge and insights
– Create natural language user experiences that provide value to users

We believe it's time for organizations to consider which applications can contribute to their goals and incorporate collaboration with fintechs.
If you're interested in learning more about how we created this approach and engaging in a conversation about the fintechs we've worked with, please feel free to contact us.

The mission of the IBM Innovation Studio is to advance the meaningful use of technology and define the path forward for organizations considering new initiatives. Our work takes place before a commercial contract is signed.

Urban Ross
IBM Innovation Studio, Studio Leader