Coverage Benchmark · KYC-2026.07
Measured Against the FFIEC Manual Itself
The denominator on this page is not ours. It is the section list published by the
FFIEC BSA/AML Examination Manual — enumerated from
bsaaml.ffiec.gov/manual, InfoBase build 2026.0423.9, retrieved 31 July 2026.
Every section is named with its FFIEC revision year, mapped to the demo lane that exercises it, and
the percentage is computed in your browser from the register below. Run
KCOV.verify() in DevTools to recompute it yourself.
Coverage · two numbers, one vocabulary
42.9%
Demonstrated
FFIEC BSA/AML Examination Manual — 19 sections in full, 4 partial, of 49. Computed in-browser via
KCOV.verify().45/45
Proven
Every case agrees. KYCRISK composites and the KYC-2026.07 policy pack were independently reimplemented in Python with Decimal arithmetic — from the published specification, not transcribed from the JavaScript — and every case executes twice. 45 of 45 agree across both implementations. Mutation-tested: injected faults were each caught, including a tampered policy threshold that consistency-checking alone would have missed.
Why two numbers. Demonstrated is what runs in
the demo today. Proven is what an evaluation harness has verified against an
independently reimplemented oracle — a higher bar than a demo lane, because it
tests the logic rather than the rendering. Across the CAIBots suite only Credit
Underwriting currently publishes a Proven figure. The other three show an em-dash,
which is the honest answer until a harness exists. Both numbers use the same
vocabulary so the four estates can be compared without translating between them.
Coverage — Computed
Full = 1 · Partial = 0.5 · excluded sections removed from the denominator
42.9% is deliberately not higher. The denominator is every section of the FFIEC manual — including 27 product, service and entity risk sections that a horizontal demonstration cannot all exercise. A vendor claiming near-complete coverage of this manual is either redefining coverage or has not met an examiner. What matters is that the sections are named, the mapping is printed, and the number recomputes in front of you.
Regulatory Requirements
Compliance Program & OFAC
Risks Associated with Money Laundering & Terrorist Financing
Nine of these were already exercised and never claimed. The demo's
entity roster does the work: an MSB check-casher is both a Non-Bank Financial Institution and a
Cash-Intensive Business; a payroll processor is a Third-Party Payment Processor originating ACH; a
foreign trade-finance bank exercises Trade Finance Activities. Those sections are credited here because
the lanes genuinely run them — not because the denominator was moved.
Roadmap
Computed at each wave from the same register
Wave 2 is the cross-estate move. Lending Activities is the section
a Credit Underwriting prospect already cares about; Nondeposit Investment Products is Investment
Research territory. A mule in Fraud Detection is a customer in KYC is a borrower in CU. No competitor
operates all four.
Excluded by Design
14 sections we will not build — and why
This list is the point of the page. Reaching 71% and naming the
14 sections a decision layer should never claim is a stronger statement than a higher percentage.
Independent Testing is a validation activity, not a lane. FBAR is a customer tax obligation. Bulk
currency shipments are logistics. Building surface for these would move a number and demonstrate
nothing.
What This Benchmark Is Not
Scope limits, stated plainly
It is not a production compliance claim. This measures what the DEMONSTRATION
exercises against the manual's sections. It says nothing about any institution's program.
It is not a model validation. Independent Testing is marked a gap because validation is out of demonstration scope — not because it is unimportant.
Section mapping is a judgement. Crediting a payroll processor to Third-Party Payment Processors, or an MSB to Cash-Intensive Businesses, is defensible from the entity type in the lane — but a BSA officer may draw those lines differently. Every mapping is printed above so it can be argued with.
One section is excluded from the denominator and named in the register: the FFIEC marks it Not Applicable for examination procedures.
It is not a model validation. Independent Testing is marked a gap because validation is out of demonstration scope — not because it is unimportant.
Section mapping is a judgement. Crediting a payroll processor to Third-Party Payment Processors, or an MSB to Cash-Intensive Businesses, is defensible from the entity type in the lane — but a BSA officer may draw those lines differently. Every mapping is printed above so it can be argued with.
One section is excluded from the denominator and named in the register: the FFIEC marks it Not Applicable for examination procedures.