Product and bizdev at a retail bank scopes remittance corridors and cross-border consumer expansion against a tight CPMI numerical set: global FPS count, cross-border-enabled subset, planning-pipeline figure, and central-bank-versus-private operator mix. Two AI failures on this regulation hit that set from opposite directions. Opus 4.7 compresses the universe to 57 and drops the operator-mix breakdown; Sonnet 4.6 holds the 70-plus headline but denies the operator-mix percentages exist. sp231115 is the primary source for the full set. A product pack built off either AI answer walks into the MD review with the wrong addressable universe and no operator-type evidence.
What the AI got wrong, and why it matters here
The two failures hit the product narrative from opposite directions, but each on its own strips a load-bearing input from the corridor expansion thesis.
Finding 1: FPS universe compressed and operator mix dropped
Opus 4.7 cited the 2025 monitoring survey at 57 (56 in one graph) operational FPS with no operator-type breakdown. sp231115 gives 70-plus operational, 14 cross-border-enabled, 24 in the planning pipeline, 40% central-bank and 35% private operator mix. A retail-bank corridor expansion thesis built on the AI answer is undersized and drops the operator-mix line.
Citation: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q010-Opus47.
Finding 2: Operator-mix denied
Sonnet 4.6 cited the 70-plus FPS headline correctly and denied that a precise central-bank-versus-private operator percentage is enumerated in the Brief 10 summary. sp231115 names 40% central-bank and 35% private. The denial removes the operator-mix evidence the partnership-rationale slide needs.
Citation: RLB-H-INT-BIS-CPMI-API-HARMONISATION-CROSS-BORDER-2024-Q010-Sonnet46.
When this hits the product calendar
Product and bizdev pulls CPMI material on four standing artefacts: the corridor expansion thesis, the partnership-rationale slide, the build-versus-license tradeoff write-up, and the board or investor update.
| Standing artefact | Where the AI risk surfaces | Failure mode |
|---|---|---|
| Corridor expansion thesis | FPS count, planning-pipeline | Finding 1 |
| Partnership-rationale slide | Operator-mix percentages | Findings 1 and 2 |
| Build-versus-license tradeoff | Operator mix as decision input | Finding 2 |
| Board or investor update | All four anchors | Findings 1 and 2 |
Aggregate impact on the team
Both failures hit the same slides in the product pack. The downside is a corridor expansion thesis built on the wrong numerical anchor or no operator-mix evidence at all.
| Risk Impact | Count | Affected findings |
|---|---|---|
| 0 |
What this team should do
Treat the FPS-landscape paragraph in any AI-drafted product deck as a controlled output. Verify against sp231115 and the latest CPMI cross-border-payments monitoring brief before the deck reaches MD or board review.
Detection patterns to add to AI-review
- FPS counts must trace to sp231115 or to a numbered CPMI brief.
- Operator-mix denial must be cross-checked against sp231115 directly.
How RLB can help
RLB tracks AI failures on d224, the CPMI brief series and the Tara Rice November 2023 speech, refreshed against live AI subjects on rotation. Retail-bank product teams can wire the catalogue into the deck-draft review step so the FPS-landscape paragraph never reaches MD or investor review without a primary-source check.