TD expands agentic AI into mortgage funding operations

TD has expanded its use of agentic AI into mortgage funding and discharge operations, extending an earlier deployment in mortgage pre-adjudication.

In the latest rollout, the tools will handle parts of document verification, connect information across different systems, identify potential discrepancies, and prepare outputs for staff review. Employees will continue to oversee the process and remain responsible for decisions.

The expansion follows the earlier pre-adjudication deployment, which cut average processing time from 15 hours to less than three minutes. TD now expects the new tools to double colleague productivity in the initial stages of funding and discharge.

Mortgage funding and discharge are often paperwork-heavy stages of home lending, involving checks across client, property, and loan records. Delays can arise when staff must gather information from several systems, compare documents, and prepare statements for final review.

TD said its funding tool produces a summary of key client and property details, allowing staff to complete an initial review in up to half the time. For mortgage discharges, the system compiles the information needed to prepare a discharge statement, which employees then check before it is issued.

Mortgage workflow

The bank is applying the technology across several points in the mortgage process rather than limiting it to a single task. This reflects a broader trend among financial institutions in Canada and elsewhere, where lenders are testing AI systems in back-office work involving repetitive checks and standardised documents.

These uses tend to focus on operational steps rather than final credit decisions. Banks have been careful to present AI as a support tool for employees, particularly in consumer finance, where documentation or approval errors can have significant consequences for borrowers.

TD framed the latest deployment in those terms, with staff remaining at the centre of decision-making. AI-generated outputs and insights are reviewed by colleagues before any action is taken.

"We're focused on using agentic AI where it can make the biggest difference for our teams and clients," said Mohit Veoli, Senior Vice President, Real Estate Secured Lending, TD. "By simplifying repetitive parts of the mortgage journey, AI can make everyday work easier and reduce friction for clients. Our colleagues remain at the centre, using the technology to augment their expertise and deliver an even better experience."

Broader AI use

Agentic AI refers to systems that can coordinate a series of tasks, analyse information from multiple sources, identify relevant points, and generate suggested outputs within a workflow. In the mortgage context, that includes reviewing documents, validating information, and preparing draft material for staff consideration.

The model is intended to reduce time spent gathering information and comparing documents, leaving employees to handle complex cases and make final judgment calls. TD also said the approach is supported by investment in staff education and skills development linked to AI use.

TD is one of North America's largest banks by assets, with 28.2 million clients and more than 14 million active mobile users across Canada and the US. The group reported $2.1 trillion in assets.

Layer 6 AI, the bank's artificial intelligence unit, has been involved in the work.

"AI is becoming an increasingly important part of how we innovate and work," said Maks Volkovs, Senior Vice President and Chief AI Scientist, Layer 6 AI at TD. "Home lending is a powerful example, evolving from a single agentic AI use case to a portfolio of capabilities that work alongside colleagues and give them more time for clients and complex decisions."