Join free with your email
Free. No spam, ever — we'll never share your email address and you can opt out at any time. Already a subscriber? Log in
Home › AI Use-Case Library › AI in document processing: 18 real deployments
Named deployments in document processing, grouped by industry.
A staffer uploaded a spreadsheet containing patient personal health information into an unapproved public AI tool for processing
Reported: 3,182 patients notified of exposed personal health data including names, admission and treatment dates, medical record numbers, account numbers, treating departments, medications and provider names
AI ambient-scribe for medical documentation and workflow automation deployed across healthcare organizations, with agentic add-ons for post-visit tasks like referrals and scheduling in development
Reported: live at more than 450 healthcare organizations, licensed through 30+ health-tech partners, projects $60-80M in revenue this year
Ardent Health clinicians use ambient AI clinical documentation as a default tool, which the hospital system frames as workforce capacity against the looming doctor shortage.
Reported: Ambient AI used in 90% of visits; documentation time cut 44%.
An AI transcription system deployed for use by doctors in Ontario generated errors and fabricated clinical content that appeared in physician notes, the provincial auditor found.
Reported: Auditor identified hallucinated details in physician notes, raising direct patient safety concerns
Northwestern rolled out the DAX ambient documentation tool for clinicians.
Reported: higher-acuity E/M billing post-DAX rollout
8,581 ambulatory clinicians across five health systems used AI-powered ambient scribes, with 1,809 adopters tracked over two years in the largest real-world ambient-scribe evaluation to date.
Reported: 13 minutes/day less in the EHR, 16 minutes less documenting, and roughly half an extra visit per week
Using the Asylum Case Summarisation (ACS) AI tool to automatically summarize asylum applicant interview transcripts for caseworkers
Reported: 45 of 203 expert-reviewed summaries failed after live rollout (22% failure rate), more than double the 9% pilot failure rate; a tribunal judge flagged possible AI-fabricated content in at least one refusal letter
Axon Draft One AI used to compose a police report documenting an investigation that involved using Flock's camera network to locate a woman who self-administered an abortion
Using AI-enabled tools to write large portions of a $435,000 Wellington City Council staffing review report
Used Claude to draft a June congressional amendment, submitting text that still contained the phrase 'Claude responded'
DOGE used ChatGPT to classify National Endowment for the Humanities grants with yes/no 'DEI-related' judgments to terminate roughly $100M in funding; a federal judge ruled the practice unconstitutional and blocked the cuts.
Reported: 143-page ruling found unconstitutional viewpoint discrimination; the model had labeled a Holocaust literature anthology as DEI-related
AI agents automating back-office paperwork for financial-advisor clients
Using automation to perform credit-market intelligence tasks including document classification, structured data extraction, and content summarization
Reported: 4M+ classifications, 1M structured data extractions, and 250K+ content summaries in 2026
AI agents handling underwriting, condition clearing, and closing reviews across multi-stage mortgage origination pipelines
Reported: revenue up 12x YoY; platform handles mortgages for lenders originating $100B+ annually
Automating loan-document checks, account reconciliation, basic coding and invoicing at Global Capability Centers in India
Automating loan-document checks, account reconciliation, basic coding and invoicing at Global Capability Centers in India
Reported: AI has cut staffing needs 30–40% in some business lines per JPMorgan CEO
Built a 214-page AI-generated knowledge wiki with 49,401 facts from a 45,047-email corpus, with every fact cited to source email, as a prototype of its company brain product
Reported: 93% of facts fully supported by source text; zero of 500 sampled facts were fundamentally wrong
deployed tender-writing agent using five structural mechanisms — machine-verified verbatim quotes with page citations, coverage accountability, gated drafting, capability auditing routing to partner slots, and explicit bracketed unknowns — to refuse fabrication rather than confabulate
Reported: on a £950,000, 133-page tender the system pulled 45 mandatory requirements and flagged 11 as unanswerable in a 5-minute draft
Every entry names the organisation and links its source. Outcome figures are quoted as reported, never estimated. Vendor announcements without a named customer are excluded. Halted and reversed deployments are kept on purpose.