Financial Services
Agent & Workflow Automation
Meridian Trust Bank
Loan-file review cut 63% with agents
that know when to stop.
Avg. loan-file review
9 weeks
1.5 h
Was
4.1 h
−63%
$2.1M
Annual cost avoided
0.62
Review threshold, conf.
9 wks
Kickoff to production
02 · The engagement
Published
01 · Challenge
What was actually broken
Not the symptom, the cause.
Meridian Trust’s commercial loan-operations team was reviewing every incoming file by hand: cross-checking financials, flagging missing documents, and routing exceptions to underwriters. Volume had grown 40% over two years without headcount keeping pace, and average file-review time had crept past three days, threatening the bank’s turnaround-time commitments to brokers.
02 · Approach
How we scoped it
And what we deliberately chose not to build first.
We mapped the review workflow end to end, including the exception paths that never made it into the official process documentation. The agent we scoped handles document completeness checks, financial cross-referencing and a first-pass risk flag; anything outside a defined confidence band routes to a human underwriter with the agent’s findings attached, not a blank file. We ran the agent in shadow mode against three weeks of live files before it touched a single real decision.
03 · Architecture
The system, in one diagram
What it connects to, what a person still reviews, where the guardrails sit.
Solution architecture
Agents
Loan files, financials
Document intake
Financials, signatures
Structured extraction
Completeness + risk checks
Workflow agent
Rules-and-model gate
Confidence ≥ 0.62
Pass · Pre-flagged file
Loan-origination platform
Else · Findings attached
Underwriter queue
Every agent action logged for audit · one-click override
A workflow agent sits between Meridian’s document intake system and its loan-origination platform, using structured extraction to pull financials, a rules-and-model hybrid for the completeness and risk checks, and a review queue for anything below the confidence threshold. All agent actions are logged for audit, and underwriters can override any flag with one click.
04 · Result
What changed
Measured in production, not in the pilot.
Within nine weeks of kickoff, the agent was handling first-pass review on 100% of incoming commercial loan files. Underwriters now start their review with a pre-flagged file instead of a blank one, and the exception queue surfaces genuinely unusual cases instead of routine missing-signature issues.
Result at a glance
Client
Meridian Trust Bank
Industry
Financial Services
Service
Agent & Workflow Automation
Time to production
9 weeks
Headline
63%
reduction in average loan-file review time
03 · What changed
In the client’s own operating numbers
63%
reduction in average loan-file review time
$2.1M
in annual processing cost avoided
9 weeks
from kickoff to first production agent
−63%
Avg. loan-file review
4.1 h
→
1.5 h
04 · In their words
Operators, not buyers
We didn't want a black box making credit decisions, and Northwind didn't build us one. The agent does the tedious cross-checking and tells us exactly why it flagged what it flagged. Our underwriters trust it because they can see its work.
Renee Castellano, VP of Loan Operations, Meridian Trust Bank
Next case study
Let’s find out
what it’s worth in production.
Have a workflow like this one? A 30-minute call is enough to tell whether it is ready to build.