Key Takeaways
- • A major mortgage lender rebuilt its 30-year mortgage system using AI-assisted coding, generating over 700,000 lines of code in the modernization effort, demonstrating significant acceleration in legacy system transformation.
- • AI-assisted pull requests merge at less than half the rate of human-authored code, with research revealing meaningfully higher rates of correctness and security issues in AI-generated lines of code, raising quality control concerns for mission-critical financial systems.
- • Over 88% of developers now use AI regularly in their workflows, reflecting industry-wide adoption of AI coding tools across mortgage technology and lending operations.
- • Industry leaders including Rocket and Blue Sage Solutions are deploying agentic AI coding agents to help mortgage lenders automate solutions and simplify technology stacks, positioning early adopters to streamline operations ahead of regulatory standardization.
- • The mortgage lending sector is shifting toward "intelligent AI" focused on outcomes, standards, and connected workflows rather than pure automation speed, with lenders recognizing that generic AI solutions cannot address complex lending requirements.
Summary
A major mortgage lender has successfully rebuilt a legacy 30-year mortgage system with over 700,000 lines of code using AI-assisted development, underscoring the technology's role in accelerating modernization of mission-critical financial infrastructure. The initiative highlights both the promise and practical challenges of deploying AI coding agents at scale within heavily regulated lending operations.
However, emerging research presents a cautionary counterpoint to this narrative of acceleration. AI-assisted pull requests merge at less than half the rate of human-authored code, according to recent developer analysis. More critically, research on AI-assisted commits has identified meaningfully higher rates of correctness and security issues compared to human-written code—a material concern for financial institutions where code reliability directly impacts borrower data, regulatory compliance, and operational integrity. This disparity suggests that while AI coding agents can generate volume rapidly, quality assurance and security review processes must be correspondingly strengthened when deploying AI-generated code in production environments.
Industry adoption remains brisk despite these quality concerns. Over 88% of developers report using AI tools regularly, and mortgage technology vendors are actively commercializing AI-assisted solutions. Rocket has deployed agentic coding approaches to modernize its RMO system, while Tavant launched a platform using AI coding agents specifically designed to help mortgage lenders build and deploy automation solutions. Blue Sage Solutions, led by executive Ash Omar, is advising lenders on how to simplify their technology stacks and adopt AI securely as these tools become standard infrastructure.
The sector is increasingly recognizing that generic AI solutions are insufficient for complex lending workflows. Industry commentary emphasizes a shift toward "intelligent AI" that prioritizes outcomes, standards, and connected workflows over pure automation speed. Rather than offering standardized 30-year fixed-rate mortgages to all borrowers, AI is being positioned to design adaptive mortgage products tailored to individual financial profiles. Lenders who move early to integrate AI into end-to-end mortgage lifecycle processes—while establishing robust code quality standards—are positioning themselves competitively as regulatory frameworks around AI in financial services continue to develop.
However, emerging research presents a cautionary counterpoint to this narrative of acceleration. AI-assisted pull requests merge at less than half the rate of human-authored code, according to recent developer analysis. More critically, research on AI-assisted commits has identified meaningfully higher rates of correctness and security issues compared to human-written code—a material concern for financial institutions where code reliability directly impacts borrower data, regulatory compliance, and operational integrity. This disparity suggests that while AI coding agents can generate volume rapidly, quality assurance and security review processes must be correspondingly strengthened when deploying AI-generated code in production environments.
Industry adoption remains brisk despite these quality concerns. Over 88% of developers report using AI tools regularly, and mortgage technology vendors are actively commercializing AI-assisted solutions. Rocket has deployed agentic coding approaches to modernize its RMO system, while Tavant launched a platform using AI coding agents specifically designed to help mortgage lenders build and deploy automation solutions. Blue Sage Solutions, led by executive Ash Omar, is advising lenders on how to simplify their technology stacks and adopt AI securely as these tools become standard infrastructure.
The sector is increasingly recognizing that generic AI solutions are insufficient for complex lending workflows. Industry commentary emphasizes a shift toward "intelligent AI" that prioritizes outcomes, standards, and connected workflows over pure automation speed. Rather than offering standardized 30-year fixed-rate mortgages to all borrowers, AI is being positioned to design adaptive mortgage products tailored to individual financial profiles. Lenders who move early to integrate AI into end-to-end mortgage lifecycle processes—while establishing robust code quality standards—are positioning themselves competitively as regulatory frameworks around AI in financial services continue to develop.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.