I recently recorded an episode of the Living Cities podcast to discuss how artificial intelligence intersects with wealth equity. While I've explored this topic in past writing, this conversation deepened my thinking about the current opportunities in this rapidly shifting political and economic landscape. One topic that we didn't get the chance to discuss in detail was specifically how to leverage AI in underwriting to optimize the efforts of community-based capital allocators.
Traditional lending practices have historically excluded millions from accessing credit and building wealth. Estimates show about 28 million Americans have no credit history and another 20 million have histories that have gone stale, meaning they are insufficient to generate a score. This systemic barrier has particularly affected communities of color, limiting their access to mortgages, business loans, and other financial opportunities.
For our work, two key leverage points for unlocking capital are 1) the flows from large financial institutions through community-based capital allocators and 2) the flows from from those local institutions to aspiring entrepreneurs and homeowners. Government initiatives that embrace AI technology and encourage coalitions have demonstrated significant success, though some programs have recently been dismantled or moved underground. The FHA's adoption of new credit scoring models and programs like Project REACh showcased AI's potential to expand credit access. By evaluating non-traditional data through AI, Project REACh enabled over 110,000 previously "credit-invisible" individuals to obtain credit cards and, through successful account management, grow their FICO scores to an average of 680.
Where the government has divested, other leaders in the field are well positioned to fill the gap. We encourage peer institutions and collaboratives to join us in these specific areas of action and advocacy:
Connect Smaller Capital Allocators with New Tech: As more new platforms emerge in economic development, we need solid evidence showing how well they boost local business ecosystems. Living Cities' strong ties with CDFIs, banks, and city leaders put us in an ideal position to test new underwriting models for small business and consumer lending—ones that expand access while strengthening capital allocators' financial performance. We can promote wider adoption by creating a stronger evidence base and showcasing proven success.
Actively Monitor Ethical AI use in Lending: Banks must treat AI underwriting with the same scrutiny as traditional credit systems. This means bias testing, scenario modeling, and fair lending audits. New tools in algorithmic fairness can help here. As regulators like the CFPB and HUD increasingly expect lenders to explain and justify their AI-driven decisions, institutions that have earned community trust must deeply understand these systems—both to advocate effectively for stakeholders and to address their own concerns.
Develop a Shared Data Infrastructure: Work with philanthropic and civic tech partners to create open infrastructure for real-time data sharing across CBCAs, including secure APIs, standardized impact metrics, and shared algorithmic testing environments. As lenders expand their credit decision data to include cash flow, bill payment history, and educational and employment records—all of which paint a fuller picture of creditworthiness—local governments can support this by providing secure access to relevant data, such as municipal utility records or public housing rent payments. This gives residents the option to share this information to strengthen their credit profiles.
Below are two models that have successfully put this vision into practice, offering both innovative technology and strategic approaches to increasing capital access:
Zest AI x Verity Credit Union
Launched in 2023, the partnership between Zest AI and Verity Credit Union, a Seattle-based CDFI with just under $1 billion in assets, has yielded swift positive outcomes. Verity's leadership identified specific biases and flawed methods that had produced unreliable lending decisions. The team integrated Zest's technology into their infrastructure, gaining detailed visibility into credit decisions and enabling automated searches for alternative repayment predictors while maintaining fair lending compliance. Clients using Zest AI's models approve 40% more borrowers from protected classes on average, with one regional bank increasing minority loan approvals by 23%—including rises of 177% for African Americans, 375% for Asian Pacific Islanders, 194% for women, and 158% for Latinos. Beyond these inclusion improvements, Zest AI's technology has enabled Verity to deliver instant decisions, boosting automated approvals by 100% for auto loans, 69% for personal loans, and 84% for credit cards. The models predict defaults 34% more accurately during economic downturns compared to traditional scoring. This success has enabled Verity to expand their loan offerings to include a microgrant program for historically undercapitalized businesses in their service area with assets under $250,000.
Uplinq
Uplinq's AI platform integrates over 10,000 data sources to analyze macroeconomic indicators, industry-specific metrics, and community data. Their machine learning models cross-validate these datasets to generate holistic risk profiles, enabling lenders to approve loans they would traditionally deny. This is achieved through a 30-50% reduction in manual review time and built-in bias mitigation that excludes racial and ethnic proxies while incorporating community resilience metrics. A recent pilot with Visa expanded the platform's use with small and mid-size business lenders in the US and Canada, resulting in a 50% reduction in underwriting operating costs, a 15x reduction in credit losses, and a 3x increase in line-of-business profitability. To date, lenders using Uplinq have approved 35-46% more loans for Black and Hispanic borrowers at 28-34% lower APRs. The platform has facilitated over $1.4 trillion in loans globally, with a particular focus on minority-owned enterprises.
Democratizing Access to Capital and Technology
For CDFIs and community banks, integrating AI-powered tools represents more than just modernizing lending practices—it's about creating a comprehensive ecosystem that supports both lenders and borrowers. Platforms like GoTackle demonstrate this dual impact: while helping CDFIs streamline operations and track impact metrics, they simultaneously empower underbanked entrepreneurs with tools to strengthen their financial positions and business operations. The platform's ability to reduce application processing time by 30-50% through automated data handling particularly benefits minority-owned businesses that have historically faced barriers in accessing traditional financial counseling.
Coalitions like Living Cities' Inclusive Capital Council—which bring together city leaders, entrepreneurs, funders, and mission-aligned investors—serve as a crucial bridge between technology and community needs. Through partnerships that prioritize both data-driven decision-making and community impact, we can move beyond creating theoretical blueprints to scaling the proven, quantifiable results that our partners have already achieved regionally across the nation.