Vertical AI Deepens in Financial Services: TD Bank Agentic Mortgages, Prudential Underwriting, and Smarsh AI Compliance

In late May 2026, vertical AI for financial services graduated from conversational assistants and back-office pilots into high-impact, production-grade deployments. Financial institutions are leveraging specialized autonomous agents to tackle complex, highly regulated workflows. Key developments include TD Bank launching an agentic AI system that slashes mortgage processing times from 15 hours to under 3 minutes, Prudential deploying a conversational underwriting assistant for independent life insurance advisors, and Smarsh partnering with Anthropic to bring Claude Enterprise under strict institutional compliance controls.

1. TD Bank Deploys Agentic AI to Fast-Track Mortgage Underwriting

On May 21, 2026, TD Bank Group (the sixth-largest bank in North America by assets) announced the production launch of its first agentic AI model. Developed by Layer 6—TD’s internal AI center of excellence—the system automates the tedious, document-heavy pre-adjudication phase for mortgages and Home Equity Lines of Credit (HELOCs).

• Performance Breakthrough: The model reduces the average pre-adjudication processing time from roughly 15 hours to under three minutes per client before a human underwriter reviews the file.
• How It Works: The autonomous agent scans borrower documents, extracts and validates financial data, calculates income, performs policy and consent checks, flags application discrepancies, and automatically compiles an executive summary memo for human underwriters.
• Strategic Financial Target: The launch serves as the flagship project of TD's enterprise-wide AI strategy. CEO Raymond Chun, who declared 2026 the "year of agentic AI at TD," has set an ambitious target of $1 billion in annual value from AI, split evenly between cost savings and revenue generation.
• Governance and Safety: To mitigate the risks of automated underwriting errors, the system operates under TD's "Trustworthy AI" framework. It does not make final credit decisions; instead, it performs discrepancy detection and routes its generated memos to human underwriters, keeping a "human-in-the-loop."
• Development History: Sandra Aziz, Senior Machine Learning Engineer at Layer 6 and Technical Product Owner of the project, noted the pioneering nature of the build: "We built where nothing else existed. Everything is new... The agentic AI agent will summarize different information from different data points and provide the right information to a human."

2. Prudential Launches "Just Ask" AI Underwriting Tool for Advisors

On May 20, 2026, multinational insurer Prudential launched Just Ask, an AI-powered conversational assistant designed to streamline administrative and underwriting workflows for its third-party advisors and active distributors across individual life insurance operations.

• Instant Underwriting Quotes: A core capability of Just Ask is its underwriting quote tool, which allows advisors to input basic client details and generate underwriting rates in minutes—a process that traditionally took several business days.
• Streamlined Workflows: The assistant answers policy questions, identifies relevant business forms, and locates product details, eliminating the need for advisors to search across multiple legacy systems.
• Early Traction: Around 85% of first-time users reported that the platform successfully resolved their queries during initial testing, with form searches and underwriting quotes emerging as the primary use cases.
• Operational Impact: Simon Berg, technology business leader for Prudential Individual Life Insurance, explained: "Our sales teams and their leaders told us they wanted faster, simpler access to information they already use every day. We built Just Ask to support those moments, especially around underwriting, where timing matters."

3. Smarsh Integrates with Claude Enterprise to Close the AI Compliance Gap

On May 21, 2026, digital communications archiving leader Smarsh announced a major integration with Anthropic’s new Claude Compliance API. This partnership enables highly regulated financial institutions to safely adopt Claude Enterprise while maintaining strict compliance with global financial regulators (e.g., SEC, FINRA).

• Full-Spectrum Archiving: The integration allows Smarsh Capture to ingest, retain, and supervise all Claude Enterprise interactions. This includes capturing text, prompts, shared files, deleted messages, archived conversations, and AI-generated artifacts.
• Risk Detection: Regulated firms can run real-time risk detection and behavioral analytics across employee AI usage, identifying potential compliance, reputational, or data-leakage risks before they escalate.
• Enterprise Control: Smarsh, which is trusted by 18 of the top 20 global banks, provides a unified view of both human and AI-generated communications. Goutam Nadella, Chief Strategy Officer at Smarsh, stated: "This integration enables our customers to adopt tools like Claude Enterprise while maintaining the transparency, control, and compliance required in regulated industries."
• The Broader Trend: This launch is part of a wider push by Anthropic, which introduced 28 security and compliance integrations (including CrowdStrike, Palo Alto Networks, and Cyera) to bring Claude Enterprise under strict institutional oversight.

Strategic Implications for the Financial Services Landscape

1. Agentic AI Drives Tangible ROI: TD Bank's 15-hour to 3-minute reduction proves that agentic AI is no longer a theoretical technology. It is a powerful competitive lever for customer acquisition and operational efficiency.
2. "Human-in-the-Loop" remains Mandatory: In highly regulated sectors like mortgage lending and life insurance, leading institutions are using AI to assist and accelerate human decision-makers rather than replace them. This design choice minimizes compliance, credit, and liability risks.
3. AI Governance is the New Standard: With Smarsh and Anthropic's partnership, the infrastructure for auditing AI-human interactions has matured. Financial institutions can no longer cite regulatory blockers as a reason to delay the enterprise-wide adoption of advanced LLMs.