PYMNTS reports that TD Bank said agentic AI reduced mortgage pre-adjudication cycle times from roughly 15 hours to 3 minutes during its May 28 earnings call. PYMNTS also reports the bank said more than 40,000 employees use internal Copilot tools and more than 7,000 engineers use AI in software development workflows. The same coverage notes U.S. proprietary credit card balances rose 18% year over year and new bank card account acquisition increased 32%, per TD executives cited by PYMNTS. Editorial analysis: Industry observers following bank deployments will watch how institutions balance throughput gains with governance, auditability and credit-risk controls.
What happened
PYMNTS reports that TD Bank said agentic AI reduced mortgage pre-adjudication cycle times from roughly 15 hours to 3 minutes on its May 28 earnings call. PYMNTS also reports TD said more than 40,000 employees are using internal Copilot tools and more than 7,000 engineers are using AI in software development workflows. PYMNTS further reports U.S. proprietary credit card balances rose 18% year over year and that TD's executives said new bank card account acquisition increased 32% following the Nordstrom conversion.
Technical details
Editorial analysis - technical context: Public coverage frames the reduction in pre-adjudication time as an application of agentic AI to orchestrate data retrieval, rule evaluation and automated decision steps that previously required manual handoffs. Industry-pattern observations note that similar pipelines typically combine retrieval from core systems, structured rule engines, and LLM-driven document parsing or decision logic, which raises practical needs for latency tuning, data lineage, and deterministic checks at decision gates.
Context and significance
Industry context: Large retail banks reporting broad internal Copilot adoption and large speedups in underwriting workflows indicate a shift from experimental pilots to production-scale automation. For practitioners, this increases emphasis on operational concerns that follow deployments at scale: throughput monitoring, explainability for credit decisions, integration with fraud engines, and change-management for regulated reporting.
What to watch
Editorial analysis: Observers should track measurable outcomes TD and peers disclose over time-error rates, appeals or reversals, regulatory inquiries, and model governance artifacts such as versioning and audit logs. Also watch whether vendors or in-house platforms publish post-deployment benchmarks that separate latency gains from end-to-end accuracy and compliance tradeoffs.
Scoring Rationale #
This is a notable production deployment by a major bank showing dramatic latency improvements, relevant to practitioners deploying AI in regulated financial workflows. It is not a frontier-model release, but it signals meaningful operational impact and governance tradeoffs.
Practice with real Banking data
90 SQL & Python problems · 15 industry datasets
[Suspicious Online TransactionsEasy](/problems/sql/suspicious-online-transactions)
[Delinquent Loans Over 30 DaysMedium](/problems/sql/delinquent-loans-over-30-days)
[Credit Card Utilization Risk ReportHard](/problems/sql/credit-card-utilization-risk-report)
250 free problems · No credit card