MT
Meridian Trust
Banking

Mirror Report — Retail Lending

Discovery run
Organization:Meridian Trust▸Business Unit:Retail Lending▸Agent:All agents (3)
As of:Run #14
View
Group by
Run #14 · Jul 29, 2026 · 09:14
Behavior Report
1,016
Sessions
4,218
Records
3/3
Agents synced
90d
Window

Across 4,218 provenance records, these agents most consistently optimize for “Approval speed & volume” and “Uniform pricing across risk tiers”. This is a description of observed behavior, not a judgment — nothing here has been compared against what leadership intends. Confirm or correct each one below to score alignment.

4 observed prioritiesNo stated priorities yet — nothing is being scored as right or wrong.Ontology v3
What these agents optimize for
Inferred from 4,218 provenance records across 1,016 sessions · ranked by evidence
1
Approval speed & volumeObserved
Credit Decisioning Agent · Retail Lending
high confidence · 92% (461/501) coverage
96% (481/501) of applicable sessions legible

Approves loans that maximize approval rate and speed-to-decision. Margin impact carries little observable weight — rate_spread barely moves the approval boundary, and collateral_value carries an approval that credit_score alone would not.

truefalseloan_approvalAPP-4498 — loan_approval true, 1.27×1.27×APP-4483 — loan_approval false, 0.96×0.96×APP-4471 — loan_approval true, 1.29×APP-4360 — loan_approval true, 1.3×
collateral_value / loan_amount ratio against loan_approval · 4 sessions resolving both
Tracksloan_amountcollateral_valueloan_approval
Is this what you intended?
2
Uniform pricing across risk tiersObserved
Pricing Exception Agent, Credit Decisioning Agent · Retail Lending
high confidence · 90% (372/413) coverage
94% (388/413) of applicable sessions legible

Prices loans to a near-constant spread regardless of risk tier. Pricing exceptions are granted without a compensating margin offset.

148 bps188 bps1234risk_tierAPP-4471 — risk_tier 3, 182 bpsAPP-4402 — risk_tier 3, 188 bpsAPP-4360 — risk_tier 4, 174 bpsAPP-4491 — risk_tier 4, 148 bpsAPP-4455 — risk_tier 3, 152 bpsAPP-4438 — risk_tier 2, 164 bpsAPP-4421 — risk_tier 1, 150 bps148 bps150 bps
rate_spread against risk_tier · 7 sessions resolving both
Tracksrate_spreadrisk_tierpricing_exception
Is this what you intended?
3
Strategic-account protectionObserved
Collections Outreach Agent · Retail Lending
high confidence · 88% (190/216) coverage
91% (197/216) of applicable sessions legible

Materially softens collections pressure for accounts flagged strategic — outreach_intensity and recovery_escalation both drop as relationship_value rises.

0.75 contacts/wk
strategic_account = true · 2 sessions
3 contacts/wk
strategic_account = false · 1 session
outreach_intensity by strategic_account · 3 sessions — too few to plot a shape
Tracksstrategic_accountrelationship_valueoutreach_intensity
Is this what you intended?
4
Thin-file cautionObserved
Credit Decisioning Agent · Retail Lending
low confidence · 74% (112/151) coverage
31% (47/151) of applicable sessions legible

Holds thin-file applicants to a visibly higher bar than their credit_score alone implies. The rule holds almost everywhere it can be seen, but only 31% of thin-file sessions resolve both concepts it depends on — the confidence is limited by the logging, not by the behavior.

Tracksthin_filecredit_score
Is this what you intended?
By domain concept
CreditworthinessObserved
Credit Decisioning Agent · 2,003 records · 501 sessions
high confidence · 92% coverage

Approvals track loan_amount & collateral_value; margin impact (rate_spread) carries little observable weight.

Pricing / marginObserved
Pricing Exception Agent, Credit Decisioning Agent · 1,240 records · 413 sessions
high confidence · 90% coverage

rate_spread sits in a 148–188bps band across every risk tier — pricing does not widen with risk.

Account valueObserved
Collections Outreach Agent · 975 records · 216 sessions
high confidence · 88% coverage

strategic_account reliably lowers outreach_intensity and suppresses recovery_escalation.