A mortgage application used to mean waiting. At TD Bank, pre-adjudication averaged roughly 15 hours before a human underwriter could open the file. That window is now under three minutes.
TD Bank Group launched its first agentic artificial intelligence (AI) model to close that gap. The system classifies borrower documents, extracts and validates financial data, calculates income, performs policy and consent checks, flags discrepancies and delivers a summary memo to underwriters before any human reviews the file.
Layer 6, TD’s internal AI research and development center, built the model alongside the bank’s technology, data, lending and risk management teams. TD serves 28.1 million clients and is the sixth largest bank in North America by assets.
Mohit Veoli, TD’s senior vice president of real estate secured lending, said the system delivers “confident decisions earlier in the homebuying process.” Luke Gee, the bank’s chief analytics and AI officer, said TD is building “a hybrid future where our colleagues and AI work together to help our clients get to a ‘yes’ faster.”
Mapped End to End
The launch covers pre-adjudication, but TD has mapped the full mortgage and home equity line of credit journey from initial document submission through to funding and plans to introduce agentic AI at each stage. The bank is also exploring agentic AI applications in other business lines.
That ambition connects to a financial target TD’s CEO Raymond Chun set at the bank’s 2025 investor day: $1 billion in annual value from AI, split evenly between cost savings and revenue improvement. Chun called 2026 the year of agentic AI at TD. The mortgage launch is the first application to go live.
Advertisement: Scroll to Continue
TD’s Trustworthy AI team oversees model deployment, evaluating each system for privacy, security, fairness, accountability and explainability before it reaches clients. The team continues to monitor models after deployment. That governance framework earned TD recognition as the best responsible AI program in North America from DataIQ in 2025.
Speed as a Competitive Lever
TD isn’t the only lender automating mortgage workflows. Better.com’s AI voice agent Betsy handled nearly 100,000 mortgage-related calls per month in 2025 and resolved 35.5% of borrower inquiries without any human involvement. The system delivered a 41% reduction in the average cost to originate and doubled lead-to-lock conversion in 2025.
The pressure is structural. At its 2025 investor day, TD identified a $40 billion opportunity to win back client mortgages held at rival lenders, with AI-driven speed cited as central to that strategy.
The risks around AI underwriting are real. Errors in income validation or policy checks can push problems downstream, surfacing late in the process when a loan collapses. National Mortgage Professional noted a pattern of borrowers approved on payroll data that doesn’t match bank deposit history, creating rework and deal failures. TD’s model includes discrepancy detection as a built-in step and routes output to a human underwriter before any decision is made.
Sandra Aziz, the senior machine learning engineer at Layer 6 who led development of the system, put the scope of the build plainly. “We built where nothing else existed,” she said. “Everything is new.”