Ophelos uses reinforcement learning and NLP to modernize debt collection

Overview

Ophelos, a debt-collection technology company, built the Ophelos Decision Engine, a fully automated ML-powered outbound communication strategy using reinforcement learning to send the best message to each customer at the right time to help resolve their debt, on the Databricks Data + AI Platform with MLflow and Feature Store. It also built OLIVE (Ophelos Linguistic Identification of Vulnerability), an NLP model served via MLflow that scans inbound customer communications to identify people in vulnerable situations. The Decision Engine drove a 25% increase in customers resolving their debts, with 88.3% of customers self-served and 89% rating their experience 4 or 5 stars.

The challenge

Debt collection agencies have traditionally employed outdated methods — such as conducting home visits, sending intimidating letters or using a rigid, one-size-fits-all communication strategy — to recover outstanding debts; these strategies are typically expensive, rarely successful in recouping payments and can hugely damage customer relations.

The solution

Ophelos builds scalable data pipelines on Databricks Data + AI Platform, powering the Ophelos Decision Engine, a fully automated ML-powered outbound communication strategy that uses reinforcement learning to send the best communication to each customer at the right time, and OLIVE, an NLP model built and served using MLflow that scans inbound communications to identify customers in vulnerable situations.

Reported business value

Using the Decision Engine, Ophelos achieved a 25% increase in customers reaching a solution, with 88.3% of customers self-served and 89% of customers rating their experience 4 or 5 stars, alongside a 67% increase in customer satisfaction.