Lenders Dream of AI, But Digital Processes Must Come First!
By Patrick Stäuble, Founder & CEO of Teylor and DLA Executive Board Member
„Artificial Intelligence is the future of lending.“ While this statement may be true and the focus of many papers and keynotes, an AI-enabled future is probably still further away than many lenders would like to admit. The hard truth is that most lending processes are still paper-based today, and going from paper-based to fully digital AI-enabled lending is a giant leap. Instead of overfocusing on AI capabilities, lenders must first lay the groundwork.
Where we stand today
According to PwC, 69 percent of German financial institutions already use some form of AI application – and 75 percent want to expand their use of AI within the next years. Their main goals are increasing efficiency (68 percent) and cutting costs (52 percent).
AI offers the financial industry a vast number of use cases. Some of the more well-known examples are Danske Bank’s use of AI and deep learning in anti-fraud, which has allegedly reduced the bank’s quota of false-positive fraud reports by 60 percent and detected 50 percent more fraud cases. Another is Chatbot Erica at Bank of America, which has already responded to around 800 million requests from more than 42 million customers.
There are many more examples, and today’s AI applications are certainly just the beginning of much more to come. However, looking at lending, particularly corporate lending, there have not yet been any groundbreaking AI applications. And there are several reasons why that will likely remain the case in the foreseeable future.
Corporate lending heavily relies on paper-based processes
Corporate lending is significantly different from retail lending: While retail customers with comparable credit profiles carry similar default risk, this is not the case in corporate lending. Two companies might have the same revenues and profits, but their business models could be entirely different, resulting in significantly different default risk. This heterogeneity in the borrower base makes assessing a corporate borrower’s creditworthiness more complex than a retail borrower’s.
This complexity is why corporate lenders stand to benefit the most from digital and AI-based processes – particularly SME lenders because smaller ticket sizes necessitate a high degree of automation. However, complexity is also the reason why it is challenging to digitalize SME lending. As a result, most lenders today still rely on manual processes and the experience of humans in the process.
Most of today’s AI use cases are built on processes that have long been digitalized. But how do you use AI in a paper-based process? The obvious answer is, you don’t! Without digital infrastructure, any application of AI remains a pipe dream. Hence, before even thinking about AI, lenders must lay the groundwork and build up digital infrastructure.
Where to start?
From our daily work with financial institutions, we know that digitalizing SME lending processes is more of an evolution than a revolution. Banks especially need to manage operational risks, adhere to strict compliance rules and regulations, and ensure compatibility with core banking systems.
To identify digitalization potentials, lenders must ask themselves where the greatest inefficiencies are and where digital processes can add the most value. From our experience, there are two areas where digitalization creates a quick Return on Investment in SME lending: loan applications and credit decisions.
According to research by Bain & Company, front office employees at banks spend 70 percent of their time on administrative tasks, such as company research, document exchange, or manually analyzing financial statements. Only 30 percent of their time is spent on customer-facing sales activities. Likewise, back-office processes are also highly inefficient: Assessing an SME’s credit risk and preparing financing contracts takes, on average, six to eight hours, and SMEs wait 10 to 30 days for a credit decision.
If banks could increase their time spent on sales and origination activities and speed up credit decisions, significant top-line potential would be unleashed. Additionally, back-office productivity gains and lower operational costs would increase lending profitability. Being able to digitalize customer interactions, analyze and qualify SMEs quickly and make accurate credit decisions frees up enormous resources in front- and back-office that can be used more efficiently for high-value tasks.
Today, it does not need AI to achieve this, simply digitalizing paper-based processes would already deliver enormous productivity gains. And perhaps even more important: Digital processes prepare financial institutions for the future! Once the first groundbreaking AI applications arrive, digital lenders will be ready to leverage the new technology.
First things first: No AI implementation without digital infrastructure
To sum up, we recommend lenders focus on digital processes before they even start thinking about AI. Going from paper-based to AI-automated lending is a giant leap and impossible for most financial institutions. Instead, the road to AI is a gradual evolution. That said, the journey will be worth it, and it’s best to start immediately.
For more information on the opportunities and risks of AI in SME lending, read the latest Teylor whitepaper Reality Check: What Can AI Do For SME Lending?
The insight can be downloaded here.
Photo Credits:
- AI: Aristal @ Pixabay
- Patrick Stäuble: Caroline Pitzke