For Provana's Loan Origination System, should we invest in developing AI-driven decision-making capabilities or enhance manual review tools to maintain human oversight in lending decisions?

Introduction

For Provana's Loan Origination System, we're faced with a critical decision: invest in AI-driven decision-making capabilities or enhance manual review tools for human oversight in lending decisions. This trade-off involves balancing automation and efficiency with human judgment and risk management. I'll analyze this through the lens of product strategy, user impact, technical feasibility, and business alignment.

Step 1

Clarifying Questions (3 minutes)

• Based on recent fintech trends, I'm thinking AI might significantly improve loan approval speed. Could you share any data on our current loan processing times and approval rates?

Why it matters: Helps quantify potential efficiency gains Expected answer: Average processing time of 3-5 days, 70% approval rate Impact on approach: Faster processing could justify AI investment if it maintains or improves approval quality

• Considering regulatory compliance, I'm assuming there are strict requirements for lending decisions. What specific regulations govern our loan origination process?

Why it matters: Ensures any solution adheres to legal requirements Expected answer: Compliance with FCRA, ECOA, and state-specific lending laws Impact on approach: May limit AI implementation or require specific explainability features

• Looking at our user segments, I'm thinking different borrower types might benefit differently from AI vs. human review. Can you provide a breakdown of our main borrower segments?

Why it matters: Helps tailor the solution to user needs Expected answer: Mix of prime, near-prime, and subprime borrowers Impact on approach: Might suggest a hybrid model with AI for straightforward cases and human review for complex ones

• Regarding our tech stack, I'm curious about our current system's integration capabilities. How flexible is our existing loan origination system for incorporating new technologies?

Why it matters: Determines feasibility and timeline for implementation Expected answer: Modular system with APIs for integration Impact on approach: Easier integration could favor quicker AI implementation

• Considering our team structure, I'm wondering about our current loan review capacity. What's the size and workload of our manual review team?

Why it matters: Helps assess the potential impact on workforce and operations Expected answer: Team of 50 reviewers, operating at 90% capacity Impact on approach: High workload might justify AI investment for efficiency, but also consider retraining needs