Which came first—the chicken or the egg? The answer often depends on who you ask. In this blog, we’re attempting to answer a far less paradoxical yet much more relevant (to us) question – What comes first—the problem or the technology? Like the age-old dilemma, the answer can vary. For us, though, there’s no confusion. Our response is always to focus on the problem first. In our opinion, technology should be a means to an end and not an end in itself.
A problem-first origin story
Ever since ChatGPT launched in November 2022, it’s been almost impossible to avoid reports of the impending ubiquity of AI and how it is set to transform everything from processing food to processing data. So, it’s unsurprising to see many companies embracing an AI-first strategy, whether it creates any actual business value or not. But here’s an interesting fact – we’ve been in the AI game long before the current wave began.
And it all started with a problem.
A problem our co-founder Joe Kurian encountered over the years working in business lending at companies such as HSBC, Wells Fargo, and GE capital – delayed risk detection. Joe and his team frequently faced the reality that, despite thorough initial due diligence, seemingly healthy loans could quickly turn into NPAs. Frustratingly, postmortem analyses of these defaults often revealed multiple early warning signs that the team had overlooked. Some of these early warnings were financial, like irregular repayment patterns, while others were non-financial, such as news reports about operational challenges. These red flags were often missed because risk reviews were conducted only quarterly at best and were focused mainly on financial data. As a result, the team would fail to reassess the company's risk profile in time, leading to avoidable portfolio losses and write-offs.
In addition, reviewing and preparing risk reports on a company is a time-consuming process, often requiring as much effort as a full credit assessment. Consequently, the team was usually more focused on high-risk companies. As Joe always puts it, “95% of risk comes from 5% of your portfolio”. Unfortunately, the composition of this 5% can change rapidly, and identifying which companies belong in this high-risk group is the veritable million-dollar question that business lenders need to answer on a day-to-day basis.
TRaiCE was born out of this problem. To cut a long story short, Joe teamed up with Sony and Geetha (our tech co-founders) to create a platform that leverages technology to fill in the gaps left by traditional monitoring. We’re blessed to have a team with a clear understanding of the problem and the right knowledge of what tech can be used to solve it.
That’s why TRaiCE uses a mix of automation and NLP techniques to ensure high-frequency, complete corporate portfolio monitoring (see table below). This way, every borrower within a portfolio gets reviewed daily. In addition, the platform also allows lenders to incorporate non-financial and other relevant industry-specific metrics into their risk assessments. Incorporating real-time data helps lenders identify and prioritize companies that require creditworthiness reviews, facilitating early risk detection and timely risk mitigation responses.
Blending the right technology with the right data provides better business value
Data and AI go hand in hand. Data shapes the technology and the technology in turn generates more valuable data. There’s plenty of data available today – 149 zettabytes (as of 2024) to be precise. Not all of it is essential for effective corporate portfolio risk management. Finding and using the right data enables more accurate and focused industry-specific assessments.
For example, fashion trend data may be relevant for evaluating a retail portfolio but typically holds no significance for a real estate account. So, we use techniques such as entity extraction and keyword tracking to quickly and more accurately identify relevant risk-and-industry-related information from the vast volumes of data available today. This way, lenders get a focused picture of risk without getting overwhelmed.
This approach helps us deliver better business value to our clients. For instance, one of our clients is a California-based community bank that wanted to increase loan volumes in the food and beverage industry by offering restaurants pre-approved credit offers. Our client’s due diligence relied on research analysts manually reviewing digital data and multiple credit reports for signs of distress. They wanted to move away from this siloed, spreadsheet-reliant approach.
With that in mind, our proprietary AI models were tuned to integrate with relevant world data and credit bureau APIs to collect information on hundreds of local restaurants, their owners, and subsidiaries. The TRaiCE platform then analyzed this information to produce time-series risk scorecards on these businesses, providing the client with a central repository of risk-related information. The result was a data-backed, lead prospecting list that was about 30% cleaner (for more details on what TRaiCE found, read our case study on the bank) and a due diligence process that is highly repeatable and more thorough.
Conclusion
Focusing on a particular technology before understanding the problem is like putting the cart before the horse - an inefficient approach that can even set you back a few steps. Prioritizing technology can also result in solutions that are unnecessarily complex and expensive. Instead, focus on the problem, and the right technology will follow.
Our team has worked to develop a risk-monitoring system designed to solve the problems business lenders face today. If this sounds interesting to you, feel free to get in touch with us at info@traice.io to get the conversation started.
Kommentare