Fintech One-On-One Roberto Salcedo of Baubap on providing short term unsecured loans in Mexico profitably
Oct 24, 2024
Roberto Salcedo, Co-founder and CEO of Baubap, leads a microlending revolution in Mexico, leveraging mobile data and AI to serve the underserved. He discusses the unique challenges of providing unsecured loans in a cash-driven economy. Discover how Baubap rapidly delivers loans through an intuitive app and their data-driven underwriting models. Roberto shares insights on improving approval rates to 40% while maintaining low default risks. He also reveals the innovative referral program that engages users, enhancing profitability and promoting financial inclusion.
AI Snips
Chapters
Transcript
Episode notes
Underwriting From Mobile Device Signals
- Baobap underwrites using mobile device data like SMS, contacts, apps, and geolocation instead of financial history.
- They feed thousands of features into 15 in-house ML models and LLMs to predict creditworthiness quickly.
Hypothesis-Driven Model Development
- Baobap began by hypothesizing daily phone interactions correlate with economic activity and repayment ability.
- They tested this by extracting founders' device data, launching a beta app, and training early AI models like BERT.
Keep Models Current And Retrain Often
- Continuously adopt improved foundation models and retrain with fresh data to boost prediction accuracy.
- Scale training data through real loan volume to turn model performance into a competitive advantage.