📺
Sponsored Content:
This was produced in partnership with our sponsor,
.
Lessons from our partner Baubap on using AI to extend credit.
Latin America represents one of the most challenging markets for credit products, particularly when serving the underserved populations.
With default rates hovering around 40% in the unbanked segment, the market often appears unattractive for many financial institutions. However, this challenge also presents a significant opportunity: reaching the 60% of individuals who are willing and able to pay but lack access to traditional credit mechanisms.
💡
Note: This data has been collected monthly since January 2022. Baubap originally randomly selected 1,500 users; since February 2024 the set expanded the sample to 3,000 monthly participants. It is a blinded randomized experiment, statistically representative with a confidence level of 99.99% and a margin of error of 3.5%.
Nearly 80% of Latin American consumers have not borrowed credit from a financial institution, creating opportunities for better data and AI in underwriting.
The Challenge: High Default Rates in a Complex Market
One of the most pressing challenges in scaling credit products in Latin America is the difficulty in assessing risk for individuals who operate outside formal financial systems. Traditional credit scoring relies heavily on credit bureaus, bank statements, and formal employment records—data points that are often missing for the region’s large informal workforce. This lack of visibility has led many financial institutions to shy away from serving this market, resulting in exclusion for millions of people.
The Opportunity: Leveraging AI to Unlock Financial Inclusion
In 2019, Baubap saw an opportunity to use AI to bridge this gap. The key insight was to leverage alternative data sources to predict repayment behavior, focusing on transactional data without relying on traditional indicators like credit bureau scores or bank statements. By analyzing a variety of data points—such as mobile device usage and behavioral patterns—Baubap began developing AI models capable of accurately predicting repayment probabilities for individuals with no formal credit history.
This shift allowed Baubap to unlock access to credit for millions of underserved consumers while maintaining a sustainable business model. The challenge lay in using AI to identify the 60% of borrowers who had the ability and willingness to repay loans, despite lacking traditional credit backgrounds.
How AI Became the Core of Baubap's Business
Baubap’s journey with AI began in April 2019, when the company started building models that could predict payment probability purely from alternative transactional data.
This was a crucial first step in scaling financial products for underserved consumers. The company’s early bet on AI paid off when it began approving loans based on real-time data from users' mobile phones, enabling loan approvals in less than 30 seconds. This gave Baubap a significant advantage in a market where speed and ease of use are critical for customer trust.
In 2020, Baubap integrated Google’s BERT model, one of the earliest large language models, into its underwriting system. This marked a major turning point for the company, allowing it to enhance its predictive capabilities even further by processing raw text data, alongside the numerical data it was already analyzing. This combination allowed Baubap to develop more sophisticated borrower profiles and reduce fraud rates to less than 0.01%.
By 2021, Baubap expanded its AI models beyond underwriting to include KYC (Know Your Customer) and anti-fraud systems. This expansion enabled the company to maintain a default rate below 10%, far outperforming industry standards, where default rates for underserved consumers are normally around 40%, and further proving the value of AI in managing risk.
Scaling with AI: Sustainable Growth and Operational Efficiency
Today, Baubap has grown into one of Latin America’s most prominent AI-powered fintechs, scaling its business efficiently while addressing the needs of underserved consumers. AI plays a critical role in the company’s ability to scale, enabling it to process over 800,000 loan applications per month, approve more than 8 million loans, and lend over $600 million since its inception.
This growth has been fueled by the company's continued focus on leveraging AI to reduce operational costs and improve efficiency. By using AI, Baubap has reduced its customer acquisition costs while maintaining a 40% contribution margin, proving that its AI-driven model is not only scalable but also profitable at scale.
The Power of AI in Extracting Value from Raw Data
One of the most significant challenges Baubap faced early on was extracting valuable insights from raw, unstructured data. Much of the data available to Baubap’s models—such as mobile usage patterns, and behavioral data—wasn’t in a format that could be easily analyzed using traditional methods. However, Baubap’s use of Large Language Models (LLMs) since 2019 has allowed it to process this raw data and generate powerful predictors of repayment behavior, outperforming many competitors and traditional financial institutions.
This ability to refine and improve AI models continuously has been crucial to Baubap's success. By investing in AI early, Baubap developed a robust pipeline capable of handling thousands of loan applications per day while providing personalized assessments for each customer.
Key Lessons for Scaling Financial Products
Baubap’s experience scaling with AI offers valuable lessons for other fintechs looking to enter underserved markets:
AI as a Long-Term Strategy: Early adoption of AI gave Baubap a head start, allowing the company to refine its models over time. This long-term investment in AI has paid off in the form of reduced default rates and increased operational efficiency.
AI Isn’t Just for the Future—It’s Driving Growth Now: While many companies are still positioning AI as a future investment, Baubap has shown that AI can drive immediate business results. The company’s AI models have enabled it to scale rapidly while maintaining strong unit economics.
Addressing the Needs of the Underserved with AI: Baubap’s success lies in its ability to use AI to serve populations that traditional financial institutions overlook. By leveraging alternative data sources and machine learning, Baubap has built a scalable, inclusive system that serves millions of underserved consumers.
