■ ROLE EXPOSURE INDEX
The evidence-based map of AI exposure across fintech roles.
Exposure measured from executed, gated tests under real controls, not model priors.
Empty ≠ 0. · No measurement, no number. · Escalation is a correct outcome.
Exposure is task share, not a layoff number. A role-level score waits for a threshold fixed in advance.
UNTESTED MAPPED PROTOCOL OUR TEST EXTERNAL STUDY
15 DIRECTIONS · 4 · 5 · 18 · 48 · TOTAL 75
ONLY OUR TEST IS MEASURED-BLUE. EXTERNAL STUDY SITS ASIDE. IT IS NOT A STEP.
■ Pack slate · 01–02 measured · 03–04 specified, no run
01 · rec.bank-ledger.v1Bank ↔ bookOUR TEST Works with controls21 posted · 15 of 36 bank lines still a desk · 2 of 2 holds correct02 · kyc.sanctions-adjudication.v1Name-screening L1OUR TEST Works0 of 25 missed · at most 13 holds · baseline missed 203 · fraud.tm-l1-triage.v1AML TM L1SPECIFIED · NO RUN04 · pay.exceptions-returns.v1Payments exceptionsSPECIFIED · NO RUN
Latest result · kyc.sanctions-adjudication.v1 · 16 Oct 2026 15:35 UTC
Name-screening L1
OUR TESTWORKS
No sanctioned party missed in three passes. At most 13 hold answers (limit 17). Hit precision at least 90.9%. The deterministic baseline failed with 2 missed sanctions.
works · 0 of 25 missed in each of 3 passes · n=69 rows · single SAMPLE book, single list version
SAMPLE · NOT A BANK BOOK · NOT HOLDOUT · 69 rows · 25 hit / 14 hold / 30 clear
First brick · rec.bank-ledger.v1
Bank ↔ ledger match
SAMPLE · NOT A REAL BANK · HARBOR LEDGER DEMO. Outcome: Works with controls. Readiness: Requires integration. Roles touched: Bookkeeping clerks, payments ops. Accountants — recon / close slice only. Candidate: 21 posted · 15 of 36 bank lines still a desk · 2 of 2 holds correct.
OUR TESTWORKS WITH CONTROLSREQUIRES INTEGRATION
■ Models · No invented eval scores. One named model has a Role Exposure pack result: claude-sonnet-5 on P02 (works, SAMPLE book, run 16 Oct 2026). Other named lab models have none. Rules baseline: ERP STP · amount + date + exact ref, ran as baseline on rec.bank-ledger.v1.
Index AETHER-FIN-2026-10-18 · Snapshot 18 Oct 2026 · Role map
01 Payments & acquiring
5 roles Payments operations / reconciliation
Chargeback & dispute operations
Acquiring / merchant operationsunmeasured
Settlement & clearingunmeasured
PSP / processor operationsunmeasured
02 Banking operations & ledger
5 roles Bookkeeping, accounting, auditing clerks
Core ledger / GL operationsunmeasured
Tellersunmeasured
New accounts clerksexternal study: FO 0.99 · not our pack
Correspondence clerksexternal study: α 95.2% · β 95.2% · not our pack
03 KYC, AML & sanctions
5 roles KYC / onboarding analyst
AML / BSA investigator
Sanctions / screening analyst
PEP / EDD analystunmeasured
Financial examinersexternal study: AIOE #2 · not our pack
04 Fraud & transaction monitoring
4 roles Transaction monitoring analyst
Alert triage (L1)
Fraud / risk ops (payments)unmeasured
Fraud rules / strategyunmeasured
05 Lending, scoring & underwriting
6 roles Loan officersexternal study: FO 0.98 · not our pack
Credit analystsexternal study: FO 0.98 · not our pack
Credit underwriter (consumer fintech)unmeasured
Loan interviewers and clerksunmeasured
Credit authorizers, checkers, clerksexternal study: FO 0.97 · AIOE #15 · not our pack
Collections operationsunmeasured
06 Risk management
5 roles Model-risk / AI-governance officerunmeasured
Market risk analystunmeasured
Credit risk analystunmeasured
Operational risk analystunmeasured
Liquidity risk analystunmeasured
07 Treasury & liquidity
4 roles Treasury / FP&A (startup)unmeasured
Cash managerunmeasured
ALM analystunmeasured
Funding / money-marketsunmeasured
08 Accounting & reporting
7 roles Accountants and auditors
Tax preparersexternal study: ζ 100% · FO 0.99 · not our pack
Tax examiners & collectorsexternal study: FO 0.93 · not our pack
Budget analystsexternal study: AIOE #5 · not our pack
Regulatory reportingunmeasured
Product controlunmeasured
Financial managersexternal study: σ² 13.0 · AIOE #13 · not our pack
09 Trading & investment operations
6 roles Financial quantitative analystsexternal study: ζ 100% · not our pack
Quant developer / strats engineerunmeasured
Financial and investment analystsunmeasured
Investment fund managersexternal study: σ² 13.0 · not our pack
Trade / middle-office opsunmeasured
Corporate actionsunmeasured
10 Wealth management
4 roles Personal financial advisorsunmeasured
Securities, commodities, and FS sales agentsunmeasured
Wealth operationsunmeasured
Suitability / advice reviewunmeasured
11 Insurance / insurtech
5 roles Insurance underwritersexternal study: FO 0.99 · not our pack
Claims adjusters, examiners, investigatorsexternal study: FO 0.98 · not our pack
Actuariesexternal study: AIOE #3 · not our pack
Policy administrationunmeasured
Insurance operationsunmeasured
12 Crypto & digital assets
5 roles Blockchain engineersexternal study: β 97.1% · α 94.1% · not our pack
Crypto / on-chain operationsunmeasured
On-call incident / crypto exchange opsunmeasured
On-chain complianceunmeasured
Custody operationsunmeasured
13 Customer operations
4 roles Customer service representativesunmeasured
Customer-support team lead (neobank)unmeasured
Complaints / ombudsunmeasured
Retention / win-backunmeasured
14 Legal, compliance & audit
4 roles Compliance officersexternal study: FO 0.08 hist. · not our pack
Internal auditunmeasured
Legal operationsunmeasured
Conduct riskunmeasured
15 Product, engineering, data & cybersecurity
6 roles Software / analytics engineers (finance employers)unmeasured
Data / analytics engineerunmeasured
Cybersecurity / AppSecunmeasured
Product managerunmeasured
Data entry keyersexternal study: FO 0.99 · not our pack
Fintech growth / lifecycle marketingunmeasured
Desk flow. Color is a verdict, not a layoff light.
Our run · P01-PUBLICWorks with controls / holdFails this pack (none yet)Protocol mapped · no runUntested · no fill