Autonomy works when the boundary is explicit.

Mortgage and HELOC pre-adjudication and application memo preparation

That preparing a mortgage or HELOC file for a credit decision requires a human to read the document package end to end - classifying documents, extracting figures, calculating income, checking consents and writing the summary - because borrower documents are too variable for a rules engine to handle reliably.

1 independent, 1 primary; publication outcomes are verified and reported.

AI-supported double reading with normal triaging and a post-read safety net (Vara MG)

Two unaided human readers plus consensus are not the only safe model for a national screening programme.

2 peer reviewed; publication outcomes are verified.

Advance Alert Monitor

A predictive alert does not need to interrupt every bedside clinician; a centralized nurse team can validate the signal and activate local clinical response only when warranted.

1 peer reviewed, 2 primary; publication outcomes are verified.

Dynamic Risk Assessment transaction monitoring

Transaction-monitoring teams do not need a large rules-generated queue to preserve risk coverage; machine-learning risk detection can narrow the queue while investigators retain consequential review.

2 primary, 1 vendor; publication outcomes are verified and reported.

CGM-driven automated insulin delivery

Bounded closed-loop control can adjust insulin while patients and clinicians retain targets and safety response.

1 independent, 1 peer reviewed; publication outcomes are verified.

Autonomous AI diabetic eye exam at point of care (ACCESS RCT; LumineticsCore/IDx-DR)

Closing the diabetic eye-screening care gap requires persuading youth to attend a separate specialist appointment. Instead, the diagnostic exam itself is brought into the routine diabetes visit and read autonomously by AI in under a minute.

3 peer reviewed, 2 primary; publication outcomes are verified.

Automated LVO detection with parallel transfer notification

Large-vessel occlusion at a non-thrombectomy primary stroke center has to be recognized and the receiving team activated through sequential imaging review and conventional communication before transfer can start.

4 peer reviewed; publication outcomes are verified.

Ambient AI draft-note generation during patient encounters

Ambient tools can draft clinical notes, while physicians remain the editors and signers.

1 peer reviewed, 1 primary; publication outcomes are verified.

Randomized deployment of DAX and Nabla ambient AI scribes

Ambient tools can draft notes in practice, but vendor-specific efficiency and error results require comparison.

1 peer reviewed; publication outcomes are verified.

AI speech neuroprosthesis for text, audio, and avatar control

Attempted-speech brain signals can produce real-time text, audio, and facial animation as a faster channel.

1 peer reviewed; publication outcomes are verified.

E-Check camera-gate, AI diagnostics, and robotic inspection

AI can flag deviations from train images; technicians validate findings and trigger work before manual repair.

1 primary; publication outcomes are reported.

Real-time AI computer-aided adenoma detection during colonoscopy

That the endoscopist must be the only visual observer capable of flagging a lesion during withdrawal.

1 peer reviewed; publication outcomes are verified.

TREWS provider-confirmed early warning

An alert only changes care when a clinician confirms it quickly.

2 peer reviewed; publication outcomes are verified.

COiN agreement extraction

Defined attributes can be extracted across annual volume while lawyers handle interpretation exceptions.

1 independent, 1 primary; publication outcomes are reported.

On-demand autonomous drone blood delivery

Remote hospitals can request blood without holding excess local stock or waiting for regional road transport.

1 peer reviewed; publication outcomes are verified.

GenAI-enabled medical claims straight-through processing

Medical claims require a human to read every document before payment. AIA instead built a three-tier GenAI stack (OCR intake, report summarization, auto-adjudication plus FWA anomaly detection) so most claims complete without human touch.

2 primary; publication outcomes are verified and reported.

AI legal-privilege screening and investigative document analysis

AI can perform first-pass privilege screening at scale while lawyers retain privilege and prosecution judgments.

Watch status: verify the cited source and deployment condition before reusing this case.

1 primary; publication outcomes are verified.

Real-time proprietary sepsis alerts

A deployed alert can add workload while missing most cases.

1 peer reviewed; publication outcomes are verified.

Prostate Pathology Quality Control

Second-read quality control is too labor-intensive to perform on 100% of cases.

1 peer reviewed; publication outcomes are verified.

Cytora headless risk digitization feeding Z-Horizon+ underwriting workbench

AI can digitize and triage submissions before underwriters make the underwriting decision.

2 independent, 1 vendor; publication outcomes are reported.

Self-learning risk-classification model for benefit applications

Fraud-risk triage can cause institutional harm without fair features, recovery rules, or challenge rights.

1 primary; publication outcomes are verified.

EY Helix full-population general-ledger analysis

Audit risk assessment can begin with population-wide analytics instead of statistical samples alone.

2 primary; publication outcomes are reported.

AFST risk-assisted investigation screening

Administrative data can standardize triage only with override and disparity review.

1 independent, 1 primary; publication outcomes are verified.

ML risk referral with human fraud decision

Fraud review can focus on model-high-risk claims if adverse decisions remain human.

1 independent, 1 primary; publication outcomes are verified.

Population-scale matching with rapid activation

Continuous population screening can connect eligible patients to a standardized activation lane.

1 independent; publication outcomes are verified.

AI-automated first-line credit analysis in Home Equity origination

First-line credit analysis and signature validation do not require human review on every home-equity file.

1 primary; publication outcomes are reported.

Decision Intelligence Pro graph-based transaction risk scoring

A sub-50-millisecond model can enrich transaction risk before the issuing bank decides approval.

Watch status: verify the cited source and deployment condition before reusing this case.

1 primary; publication outcomes are reported.

Near-real-time machine-learning prioritization of potentially fraudulent Treasury checks

That potentially fraudulent checks must be found through slower post-payment reviews rather than prioritized in near real time.

2 primary; publication outcomes are reported.

SPOT continuous sepsis surveillance

Sepsis surveillance should follow patient data continuously rather than wait for nursing rounds or shift changes.

Watch status: verify the cited source and deployment condition before reusing this case.

3 primary; publication outcomes are reported.

GenAI Quick Quote preliminary assessments

A licensed underwriter need not assemble every preliminary before an agent gets a non-binding indication.

2 primary; publication outcomes are reported.