What TD Bank’s $1B AI Bet Teaches Community Lenders About Prescreen

TD Bank has wagered big on artificial intelligence—2,500 dedicated specialists, a proprietary foundation model called TD AI Prism, and a public target of $1 billion in annual value from AI applications.^[1] For community bank and credit union executives watching from the sidelines, those numbers can feel more alienating than instructive.

But buried in TD’s strategy is a framework that translates directly to institutions a fraction of its size—particularly when it comes to loan growth through prescreen marketing. The lesson isn’t about matching their budget. It’s about matching their discipline.

The Trust Equation TD Is Solving

According to Ted Paris, SVP and head of analytics, intelligence and AI for TD Bank U.S., the bank’s entire AI apparatus is oriented around one priority: “The most important thing that we’re all collectively solving for, that we all have a mutual interest in, is the trust factor.”^[1]

That trust framework manifests in specific operational guardrails. TD employees cannot download unauthorized models onto company systems. Customer-facing staff are prohibited from using generic tools like ChatGPT for client interactions. Models are purpose-built and query-limited—a tool designed to help contact center staff answer customer questions cannot be repurposed for unrelated tasks.^[1]

For community FIs, this isn’t theoretical risk management. It’s a compliance blueprint. When AI touches lending decisions—especially prescreen campaigns using bureau data—the guardrails matter as much as the algorithms.

Where the Revenue Growth Actually Lives

TD’s projected $1 billion in AI value splits evenly: half from cost reduction, half from revenue growth.^[1] Community institutions often fixate on the efficiency side because it’s easier to measure. But the revenue half is where prescreen marketing creates asymmetric returns.

Consider the math. The average community bank holds approximately $330 million in assets, according to ICBA data.^[2] A prescreen campaign that generates even $5 million in new loan volume at a 3% net interest margin produces $150,000 in annual spread income—before accounting for relationship deepening or deposit growth from new borrowers.

TD is deploying AI to “predict customer needs and personalize their experiences” at scale.^[1] Prescreen marketing, executed with modern targeting, accomplishes the same goal for community lenders: identifying households in your footprint who are creditworthy, likely to respond, and underserved by their current provider.

Why Prescreen Is the Ideal “Human in the Loop” Use Case

TD’s Raymond Chun called 2026 “the year of agentic AI”—systems that can take autonomous action across bank operations.^[1] But even TD maintains human oversight for customer-facing applications, recognizing that regulators and customers alike are watching how banks deploy these tools.

Prescreen marketing is structurally suited to this trust-first approach:

• Bureau data provides the foundation. Credit attributes from Experian establish objective qualification criteria—no black-box decisioning required.
• Firm offers create regulatory clarity. FCRA-compliant prescreened offers come with established legal frameworks that predate AI entirely, reducing novel compliance risk.
• Loan officers own the relationship. The AI identifies the opportunity; your people close it. That’s the human-in-the-loop model TD is engineering at massive scale, available to community FIs without building anything proprietary.

Fair lending scrutiny intensifies when algorithms influence credit access. Prescreen sidesteps the riskiest territory by using regulated data sources and transparent offer structures while still capturing the targeting precision that drives response rates.

The Infrastructure Gap Is Smaller Than You Think

TD acquired Layer 6, a leading AI developer, in 2018 and operates a dedicated center in New York City.^[1] Community institutions will never replicate that infrastructure—nor should they try.

The strategic insight is that TD’s guardrails exist precisely because AI is powerful enough to cause harm. Those same guardrails—purpose-limited models, compliance validation, data privacy controls—should be non-negotiable criteria when evaluating any vendor-provided AI solution.

Questions to ask before deploying AI-assisted prescreen:

• Can the model be used for purposes beyond its stated function?
• How is fair lending compliance validated before offers are generated?
• Where does member or customer data reside, and who can access it?
• What audit trail exists for offer decisions?

Paris notes that TD maintains protections including “evaluation of model risk, assurance of compliance and fair banking, maintenance of customer data privacy, and keeping bank data insulated from external exposure.”^[1] Any vendor serving community FIs should meet the same standard.

The Community FI Advantage TD Cannot Replicate

TD’s scale enables experimentation. But scale also creates distance. When a $1.9 trillion institution deploys AI to personalize experiences, it’s engineering around an inherent relationship deficit.

Community banks and credit unions start with the relationship advantage TD is spending billions to simulate. Your loan officers know borrowers by name. Your branches serve neighbors. Your credit decisions reflect community context that no model fully captures.

Prescreen marketing amplifies that advantage rather than replacing it. AI-driven targeting identifies who in your market is creditworthy and likely receptive. Your team converts that intelligence into relationships that megabanks cannot match—not because they lack the technology, but because they lack the proximity.

The billion-dollar lesson from TD isn’t that community FIs need bigger AI budgets. It’s that trust-first frameworks and human oversight turn artificial intelligence into authentic growth. Prescreen, deployed with the right guardrails, is how community lenders capture enterprise-level precision while preserving the relationship model that remains their most defensible competitive position.

Learn more here:

References

1. The Financial Brand: How AI is Permeating Work Throughout TD Bank, and Lessons Learned
2. ICBA: Community Banking Facts and Statistics