It was my first time at Fintech Nexus (formerly Lendit), and there were many highlights: the always eloquent and thought-provoking Harris Act™ (Superintendent Adrienne Harris and Bain Capital’s Matt Harris), the endless swag, Cross River’s custom sand sculpture…the list goes on.

Overall, the content spoke to the effects of the pandemic, the market uncertainty, and the increasing ubiquity of fintech in Americans’ everyday lives. Whether covering digital identity, hourly workers, or fintech influencers (should they be regulated?! Fascinating question I hadn’t considered), one key theme kept emerging: data.
Is that surprising? Not at all. But I did notice continued evolution in rhetoric from years ago: conversations have progressed significantly from the “Data is the New Gold” credo and centered more so on the challenges of creating value from data: Data Ops, the Data + Human relationship, even how we define it. We’re talking less about what technology we use and more about how we use it (which is a daily topic of conversation for us at Stratyfy).
Here are some of my key takeaways:
1) There is opportunity in the data mess.
In “Finance Without Borders: Enabling Financial Health for New Immigrants,” leaders from American Express and Nova Credit highlighted that the challenge is not that consumers who are immigrants are high risk – in fact, they’re often considered super prime in their home countries. The challenge is that there’s a process gap in how we use the data when an individual immigrates to the US.
At least 60 million consumers are locked out of mainstream credit in the US, even if they aren’t the highest risk borrowers, primarily because of this gap. What’s more, popular predictive alternative data sources, such as utilities and rent, are highly fragmented and lack uniformity, making their application more difficult. According to the FinRegLabs’s 2021 report “Utility, Telecommunications, and Rental Data in Underwriting Credit,” it’s estimated that only 2-5% of consumers who make such payments have that data in their credit bureau files today.
Cash flow data came up yet again as an increasingly proven option in addressing the limitations of existing credit risk data practices in “Deep Dive into Bank Data and Cash Flow Underwriting.” But again, how this data is used has its challenges. For instance, one speaker reminded us that for the nonstandard workforce (ie, hourly or gig workers), there is no centralized data source akin to that of W-2 earners. Therefore, it’s difficult to verify income and get useful data to better serve them – which is especially critical when many within these populations rely on public benefits and other resources that require verifying this data.
But there’s hope – because companies like Nova Credit, Esusu, and others are finding opportunities for innovation within the mess. Co-Founder and CEO of Prism Data Jason Gross, for example, shared the story of Prism’s evolution out of Petal Card, which was developed as a solution to a huge roadblock: the messy, complicated world of financial transaction data, ie a consumer’s banking history.
And at Stratyfy, we’re finding opportunities in the mess, too: our latest product, UnCover, allows financial institutions and fintechs to optimize model performance while minimizing the time and resources wasted on feature engineering. Data scientists spend a tremendous amount of time, nearly 80% in some cases, preparing and managing data for analysis – and that just shouldn’t be the case.
2) Data means different things to different people.
The term “business intelligence” was believed to be coined as early as 1865 to describe the use of data analysis to drive competitive advantage in business. Since then, data analytics have evolved considerably – and the Closing Wealth and Opportunity Gaps for Hourly Workers & Communities of Color panel reminded us that even the core meaning of data is fluid.
I’ve already alluded to this in my previous point with the fallibility of credit scores – for many, this number is an extremely limited (and limiting) view on their financial strength. And from redlining to predatory lending, the structural racism embedded in data and models may very well be THE most pressing challenge fintech must address (check out Data for Black Lives for a more complete picture).
Additionally, panelists spoke to incentives: what is our incentive structure for providing data? For instance, many consumers will gladly offer up positive payment history, but they may be more cautious when it comes to full file reporting – which is understandable but can present challenges from a predictive perspective. But because data has often been used as a weapon of oppression, many people, especially marginalized communities, lack trust in the system.
Since our beginnings at Stratyfy we’ve always featured bias mitigation in our platform – which has only been achievable through the first-of-its-kind ML transparency. With increased demand and regulation, we now offer bias detection and mitigation as a standalone tool – UnBias™– because we know that in order to change the financial landscape for the better, we can’t continue to bake bias into our models because of the bias inherent in our data.
3) Data is a science…and an art.
What I mean by that is this – there is still a human in the machine. In “Removing Bias in Lending for a More Equitable Financial System”, Ulysses Smith of Blend reminded us that we need to be creative about the data sets we use. In “Reimagining the Human Touch with AI”, Murli Buluswar, Head of Analytics, US Consumer at Citi Bank also spoke to the potential in untapped data exhaust to drive nonlinear value, but stressed that financial institutions must address the missing links between data science and commercialization teams (another opportunity in the gap between human and machine – and one that Stratyfy’s transparent platform is actively solving for). As one of his top 5 considerations when deploying ML, he called out the risk of both human and model bias and asked, “what are the financial implications of end to end automation (taking the expert out of the loop) from a risk and cost savings perspective?” Panelists spoke to the key point that oftentimes it doesn’t matter if AI or humans (or both) are driving decisions – it just matters that the decisions are accurate, efficient, and fair. And from our perspective at Stratyfy, we take this a step further to ask: how can you ensure this without transparency and understanding of how decisions are being made?
It’s also not just about data from the ingestion and processing angle – it’s also about how the data is generated and used. In “Closing Wealth and Opportunity Gaps for Hourly Workers & Communities of Color,” panelists stressed the importance of building products with the communities you’re serving in order to both build a better product and build trust. In Wednesday morning’s keynote, Swati Bhatia, Head of Marcus at Goldman Sachs, reiterated this point, saying that amidst a highly fragmented, increasingly digital financial landscape, financial institutions must build engagement in order to create trust.

So in closing, we still haven’t figured everything out – and that’s okay. But I appreciated the overarching sentiment at Fintech Nexus that instead of just looking ahead to the next shiny thing, let’s figure out how best to use–and improve–what we’ve built and the technology we have in order to meet the needs of those traditionally overlooked by our industry.
Leaving Fintech Nexus, I couldn’t help but reflect on what we’ve been building since 2017 here at Stratyfy. We’ve been focused on truly transparent AI even before it was sexy. Why? Because transparency doesn’t shy away from the gaps, the problems, the opportunities. Transparency builds equity, trust, and communication – not to mention better accuracy, efficiency, and profitability.
We know there are many opportunities for AI to radically improve peoples’ lives. There are opportunities in the gaps, in the blockers, in the friction between human and machine – let’s take advantage of them to make our financial system stronger and fairer.