Aicoll Improves Loan Default Prediction
using machine learning models in MySQL HeatWave

"Without a doubt, Oracle has helped us sell our loan default solutions to credit institutions throughout Colombia and beyond thanks to the automated machine learning engine within MySQL HeatWave, and to the high availability and scalability of Oracle Cloud Infrastructure."

Yelitza Romero
Artificial Intelligence Collection System S.A.S

Colombian leader in loan default prediction solves customer debt recovery pains by innovating machine learning models in MySQL HeatWave.

Introduction & Background

Aicoll - Artificial Intelligence Collection System - provides AI solutions for managing the risk of customers defaulting on loans. Its dashboards and business intelligence tools give customers a 360-degree view of their users, with data collected from databases, spreadsheets, web services, Know Your Customer (KYC) sources, payments gateways, Application Programming Interfaces (APIs), enterprise resource planning (ERP) systems, risk scoring models, social media, and more.

A default on debt occurs when a borrower fails to make payments, typically missing multiple payments for several weeks or months. Banks are typically equipped with the technology to help predict and prevent defaulting, but the same cannot be said for the non-banking credit and loan sector.

In Colombia, according to analytics and strategy management leader Aicoll, 60% of institutions providing loans in the retail, microservices, or cooperative finance sectors are not equipped with this technology. More than half of their senior management time is spent on formulating debt collection strategies based on data collected manually, which is often unstructured and inaccurate.

Business Challenges & Goals

Aicoll needed an automated way to massage all the data into Machine Learning (ML) models that would allow its customers to create predictions covering the credit life cycle from loan origination to debt default.

Previously using the MySQL database on-premises, requiring data scientists to extract data for Online Analytical Processing (OLAP), it discovered the breakthrough MySQL HeatWave Cloud service, the first database product to handle transactional and data analytics in one database.

Business Results & Metrics

By integrating MySQL Heatwave into its products, Aicoll provides real-time analytics and business intelligence reports to support strategic decision-making at each stage of the credit cycle, allowing customers to assess the economic benefits of implementing specific portfolio recovery measures.

MySQL HeatWave reduced the time to build machine learning (ML) models from three months to one week, enabling Aicoll to quickly meet business requirements from new customers concerned with loan defaults.

Moving to MySQL Heatwave's data cleansing, data enrichment, and data structuring has enabled Aicoll developers to accomplish in minutes the generation of default risk data which previously took an entire day.

As a result, Aicoll's customers can now execute ML-generated default predictions for thousands, if not millions of users in real-time, without having to extract data outside of the MySQL database.

Moreover, the MySQL HeatWave AI/ML-powered platform has enabled the Aicoll operations team to automatically create new machine learning models generated from statistical algorithms originating from loan origination and default data.

MySQL HeatWave's single database for online transaction processing (OLTP), online analytical processing (OLAP), and ML has resulted in up to 20% liquidity improvement across client portfolios, as credit institutions adopt MySQL HeatWave's predictive models.

By leveraging an ML-powered database hosted on Oracle Cloud Infrastructure (OCI), Aicoll guarantees its customers maximum uptime, data loss prevention, and automatic scalability for expanding their businesses, handling from 50,000 to one million users and more smoothly.

In addition to high availability and on-demand scaling, OCI provides the highest security for sensitive financial data.

Aicoll customers in Colombia - often also operating in Mexico, Brazil, Peru, and Costa Rica - see the partnership with Oracle, a major player in each of these countries, as very positive for stimulating their business growth.

Why MySQL HeatWave?

As a start-up in 2019, Aicoll selected MySQL Database Service as a low-cost solution guaranteeing business continuity.

With the launch of MySQL HeatWave, the company identified automated machine learning as a major benefit, along with higher availability, scalability, security, and support. MySQL Support facilitated the migration to HeatWave and the training of Aicoll's operational staff.

"The support delivered by the MySQL Team has been, and continues to be, first-rate. They work with us hand in hand to build out our solutions and enabling us to expand throughout Latin America," said Yelitza Romero.

Next Steps

The solution is supporting Aicoll's ambition to become Latin America's largest financial technology hub. The company's immediate strategy is to enter the Mexican market, bolstered by the opening in 2022 of an OCI region in Queretaro, Mexico that will complement the US-based data center currently used for Colombia operations.

Aicoll's technology strategy is to follow Oracle's technology roadmap. "With the outstanding support we get from MySQL and its cutting-edge solutions, we will at all times look to Oracle when assessing future requirements that will bring more value to our customers," concludes Yelitza Romero.