The Use Cases Surface.
Section Four · The Use Cases Surface
Eight horizontal use cases, every industry where they apply.
The use case is the unit of operational AI work. The Use Case Coverage Matrix shows the same five-stage positioning applied across eight horizontal use cases and twenty industries. Some use cases are present and mature across nearly every industry. Others concentrate in a handful of industries. Reading where the use case is and is not applicable matters as much as reading where it is mature.
| Use Case | Bank | FinT | Telc | RE-D | RE-M | Rtl | F&B | O&G | Hsp | Lgs | Ship | Cou | BPO | Auto | Phr | HC | Mng | R&U | Gov | Mall |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| KYC / AML | 4 | 4 | 3 | 2 | 2 | · | · | · | 3 | · | 3 | · | 4 | 2 | 3 | 3 | · | · | 5 | · |
| Inventory Management | · | · | 3 | 2 | 3 | 5 | 4 | 4 | 4 | 5 | 4 | 4 | · | 4 | 4 | 4 | 4 | 4 | 3 | 3 |
| Predictive Maintenance | · | · | 4 | 3 | 4 | 3 | 3 | 5 | 3 | 4 | 4 | 3 | · | 4 | 4 | 3 | 5 | 5 | 3 | 4 |
| Fraud Detection | 5 | 5 | 4 | 2 | 3 | 4 | 3 | · | 4 | 3 | · | 3 | 4 | 3 | · | 4 | · | 5 | 4 | 4 |
| GenAI Customer Service | 3 | 4 | 3 | 3 | 3 | 4 | 2 | 2 | 3 | 3 | 2 | 4 | 3 | 3 | 3 | 2 | 3 | 4 | 4 | 3 |
| Document AI / IDP | 4 | 4 | 3 | 3 | 4 | 4 | 3 | 3 | 3 | 3 | 3 | 3 | 4 | 3 | 3 | 4 | 3 | 3 | 3 | 3 |
| Computer Vision for Operations | 3 | 4 | 3 | 3 | 4 | 3 | 3 | 4 | 3 | 4 | 3 | 4 | · | 5 | 4 | 5 | 5 | 4 | 5 | 5 |
| Pricing Optimization | 4 | 4 | 4 | 4 | 4 | 4 | 4 | 3 | 5 | 3 | 3 | 3 | · | 3 | 3 | 3 | 3 | 3 | · | 3 |
Column abbreviations: Bank = Banking, FinT = Fintech, Telc = Telco, RE-D = Real Estate Development, RE-M = Real Estate Management, Rtl = Retail, F&B = F&B, O&G = Oil and Gas, Hsp = Hospitality, Lgs = Logistics, Ship = Shipping, Cou = Courier, BPO = Outsource Call Center / BPO, Auto = Automotive, Phr = Pharma, HC = Healthcare / Medtech, Mng = Mining and Metals, R&U = Renewables and Utilities, Gov = Government / Public Sector, Mall = Mall Management. · symbol indicates not applicable.
Cross-industry use cases versus concentrated use cases
Three of the eight use cases are cross-industry in the strict sense: they apply meaningfully in more than fifteen of the twenty industries surveyed, and they sit at Stage 3 or higher across nearly all of them. Know Your Customer and Anti Money Laundering applies in eleven industries and reaches Stage 4 or 5 in five of them. Document AI and Intelligent Document Processing applies in every one of the twenty industries and reaches Stage 4 in seven. Predictive Maintenance applies in eighteen industries and reaches Stage 4 or 5 in eleven. These three are the structural cross-industry surfaces in the workstream. Generative AI Customer Service applies in every industry but reaches Stage 4 in only three; the Trough exposure is heaviest in this row and the cautionary cohort is densest on this surface.
Two of the eight use cases are concentrated rather than cross-industry. Pricing Optimization reaches Stage 5 only in Hospitality — on the back of forty years of airline and hotel revenue management — and reaches Stage 4 in fintech, banking, real estate, retail, F&B, and telco. The concentration reflects an underlying truth: pricing optimization requires a demand curve, a yield surface, or both, and not every industry produces them. Computer Vision for Operations reaches Stage 5 in five industries: Automotive, Healthcare and Medtech, Mining and Metals, Government and Public Sector, and Mall Management, plus a Slope-of-Enlightenment reading in Renewables and Utilities. The concentration is structural: computer vision needs a visual artifact — a manufactured product, a radiograph, a haul truck, a face for identity matching, a foot-traffic stream — and the industries that produce those artifacts run the deepest deployments.
The KYC and AML row
Banking and Fintech sit at Stage 4 on KYC and AML. The integrated stack from document capture through biometric liveness through entity resolution through ongoing transaction monitoring is in production at tier-1 institutions. Vendors with banking-grade reference customers include NICE Actimize, Oracle Financial Crime and Compliance, SAS, FICO Tonbeller, ComplyAdvantage, Quantexa, Sayari, and Sumsub. Government and Public Sector sits at Stage 5 in jurisdictions with digital-identity backbones — Aadhaar in India, Estonia, the UAE Pass and Saudi Absher stacks — where the citizen-side of the KYC primitive runs at population scale. The AML side carries the failure-mode exposure documented in Section Seven of this report. False-positive saturation is the binding constraint and the regulatory enforcement pattern is now active.
The Document AI row
The Document AI row is the deepest cross-industry coverage in the workstream. The technology lineage runs from ABBYY in 1989 through the IDP wave of 2018-2023 through the GenAI-augmented IDP of 2024-2026. Banking, Fintech, Retail, Outsource Call Center and BPO, Real Estate Management, and Healthcare and Medtech sit at Stage 4. The remaining fourteen industries sit at Stage 3, mostly because GenAI-augmented IDP has been layered onto pre-existing rule-based document workflows in pilot rather than fully integrated. The MEA Arabic-language overlay on document AI is high-severity across customs, government archives, ZATCA e-invoicing, and Sharia-compliant financial reporting; the gap is one of the workstream's most consistent findings.
The Predictive Maintenance row
Predictive Maintenance reaches Stage 5 in Oil and Gas, Mining and Metals, and Renewables and Utilities. Rotating equipment, refinery process control, mineral processing, wind turbine condition monitoring, and solar PV soiling detection are all running at industrial scale. The deeper lineage is the vibration-analysis discipline, where AVEVA, Aspen, AspenTech, Senseye (Siemens), and OSIsoft (now AVEVA PI) have built the production stack across forty years. The cautionary case — GE Power Predix, a central component of the multi-billion-dollar Power-segment impairments across 2018 and 2019 — sits on the platform-overinvestment failure mode rather than on the underlying domain. PdM works. The platform thesis that tried to bypass the specialist vendors did not.
The Generative AI Customer Service row and the Trough exposure
The Generative AI Customer Service row carries the largest Trough surface in the matrix. Stage 2 readings appear in F&B, Oil and Gas, Shipping, and Healthcare and Medtech. The Stage 2 readings are not arbitrary — each one anchors a specific cautionary case. F&B carries the McDonald's-IBM drive-thru voice retreat. Oil and Gas carries the GenAI-for-technical-documentation Peak-to-Trough cohort. Shipping carries the autonomous-vessel and MASS Level 3 to 4 Peak. Healthcare and Medtech carries the Babylon Health collapse and the patient-facing GenAI chat retreat. The Stage 3 readings — Banking, Telco, Real Estate, Logistics, BPO, Automotive, Pharma, Mining and Metals, Mall Management, and Hospitality — carry concentrated success cases at named operators but no industry-wide production read.
The Stage 4 readings — Fintech (Klarna at narrowed scope), Retail (named e-commerce operators), Courier (operations chat), Renewables and Utilities (utility customer service), and Government and Public Sector (citizen-service chatbots with the NYC MyCity cautionary frame) — reflect industries where GenAI customer service has crossed into measurable production at scale. None of these reaches Stage 5. The Plateau on generative AI customer service is not yet established in any industry surveyed.