← All use cases

TD's agentic AI cuts mortgage pre-adjudication from ~15 hours to under 3 minutes

TD Bank Group · Toronto, ON · Banking · Roles: Operations, Risk & Underwriting

Money & BillsProductivity

What was done

TD's Layer 6 team built autonomous agents that classify client documents, extract information, calculate income, check figures against policy, run consent checks and flag discrepancies, then write a summary memo for the human underwriter.

Outcome

✓ 15 hrs → <3 min

TD says early results cut the process from an average of 15 hours to under three minutes, with more accurate results. Underwriters still make the decision.

Evidence: Company press release. Figures are as reported by the source and have not been independently audited by AgentsNorth.

Agents & tools used

Layer 6 agentic AI (in-house)

Prompt Reconstructed by AgentsNorth

The source didn't publish its prompt. This is our reconstruction of a starting prompt for a similar workflow, not the company's actual prompt.

You are preparing a pre-adjudication memo for a human underwriter. From the attached application documents: classify each document, extract income and liabilities, recompute income, check it against the policy thresholds provided, note any missing consent, and list discrepancies. Output a one-page memo. Do not make an approval recommendation.

Source

TD Stories · published 2026-05-21 · link checked 2026-10-15