report.object
AI agents for patient access and clinical documentation
Expert-reviewed from public records, platform data, primary sources, and private research beyond AI search. Use /api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/sample before purchase, then download PDF, JSON, or Markdown after checkout.
page.markdown
AI agents for patient access and clinical documentation.md
# AI agents for patient access and clinical documentation
Patient-access and documentation agents should be approved by owner and record path: PHI lifecycle, EHR permissions, recording consent, retention, clinician review and autonomous-use limits.
## Object
| field | value |
| ---------------- | ---------------------------------------------------------------------- |
| report_id | RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05 |
| schema_version | 2026-05-28 |
| last_updated_at | 2026-06-17 |
| category | AGENTS |
| sector | Healthcare delivery |
| region | USA |
| confidence_score | 0.86 |
| source_count | 72 |
| evidence_count | 72 |
## Endpoints
| action | endpoint |
| ------------ | ------------------------------------------------------------------------------------------------- |
| summary | /api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05 |
| sample | /api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/sample |
| schema | /api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/schema |
| quote | /api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/quote |
| purchase | /checkout/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05?view=agent |
| versions | /api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/versions |
| changelog | /api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/changelog |
| monitor | /api/v1/monitors?report_id=RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05 |
| mcp_resource | colophon://reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/sample |
## Coverage
| field | value |
| -------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| category | AGENTS |
| sector | Healthcare delivery |
| region | USA |
| topics | healthcare AI agents, patient access, clinical documentation, ambient scribing, EHR writeback, HIPAA, BAA, patient consent, recording consent, clinical review, payer administration, refill workflows, eligibility, intake, coding support, patient engagement, prior authorisation, consumer health data, TCPA, right of access, de-identification, liability |
| source_classes | vendor_docs, customer_case_study, implementation_docs, security_docs, safety_docs, privacy_terms, marketplace_listing, pricing_page, partner_case_study, regulator, professional_guidance, policy_framework, terms, state_law, independent_paper, independent_benchmark |
| update_cadence | Weekly |
## Sample JSON
```json
{
"report_id": "RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05",
"slug": "ai-agents-patient-access-clinical-documentation",
"schema_version": "2026-05-28",
"title": "AI agents for patient access and clinical documentation",
"abstract": "Patient-access and documentation agents should be approved by owner and record path: PHI lifecycle, EHR permissions, recording consent, retention, clinician review and autonomous-use limits.",
"category": "AGENTS",
"sector": "Healthcare delivery",
"region": "USA",
"topics": [
"healthcare AI agents",
"patient access",
"clinical documentation",
"ambient scribing",
"EHR writeback",
"HIPAA",
"BAA",
"patient consent",
"recording consent",
"clinical review",
"payer administration",
"refill workflows",
"eligibility",
"intake",
"coding support",
"patient engagement",
"prior authorisation",
"consumer health data",
"TCPA",
"right of access",
"de-identification",
"liability"
],
"last_updated_at": "2026-06-17T00:00:00Z",
"source_count": 72,
"evidence_count": 72,
"confidence_score": 0.86,
"web_url": "/r/ai-agents-patient-access-clinical-documentation/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05",
"sample_endpoint": "/api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/sample",
"inventory_endpoint": "/api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/inventory",
"full_endpoint": "/api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/full",
"schema_endpoint": "/api/v1/reports/RPT_HEALTHCARE_AI_AGENTS_PATIENT_ACCESS_CLINICAL_DOCUMENTATION_2026_05/schema",
"redaction": {
"status": "sample",
"paid_sections_redacted": true,
"preview_limits": [
"summary",
"table_of_contents",
"evidence_profile"
]
},
"coverage": {
"category": "AGENTS",
"sector": "Healthcare delivery",
"region": "USA",
"topics": [
"healthcare AI agents",
"patient access",
"clinical documentation",
"ambient scribing",
"EHR writeback",
"HIPAA",
"BAA",
"patient consent",
"recording consent",
"clinical review",
"payer administration",
"refill workflows",
"eligibility",
"intake",
"coding support",
"patient engagement",
"prior authorisation",
"consumer health data",
"TCPA",
"right of access",
"de-identification",
"liability"
],
"source_classes": [
"vendor_docs",
"customer_case_study",
"implementation_docs",
"security_docs",
"safety_docs",
"privacy_terms",
"marketplace_listing",
"pricing_page",
"partner_case_study",
"regulator",
"professional_guidance",
"policy_framework",
"terms",
"state_law",
"independent_paper",
"independent_benchmark"
]
},
"confidence": {
"overall_confidence": 0.86,
"source_quality": 0.88,
"source_recency": 0.93,
"source_density": 0.89,
"claim_verifiability": 0.85,
"geographic_coverage": 0.78,
"known_gaps": [
"Public healthcare-agent proof is strongest for named deployments and weaker for independent workflow benchmarks.",
"Ambient documentation vendors publish improving enterprise evidence, but comparable clinician edit burden, incident rates and speciality-specific performance remain uneven.",
"Public benchmark and note-quality evidence has improved, but it still argues for assistive approval before autonomous clinical action.",
"Patient-access ROI claims often omit denominator, baseline, no-show adjustment, repeat contact, staff rework and audit method.",
"BAA, subprocessor, retention, model-training, designated-record-set, recording-consent and EHR-writeback terms frequently require NDA or contract review.",
"Prior-authorisation AI evidence is moving quickly, but denial-reason, appeal, human-review and patient-harm evidence remains difficult to compare publicly.",
"US regulatory evidence is strong; non-US deployment rules need local privacy, medical-device, data-residency and clinical-governance review."
]
},
"document": {
"object": "report_document",
"schemaVersion": "2026-05-28",
"title": "AI agents for patient access and clinical documentation",
"subtitle": "Approve the owner, PHI lifecycle and record path before the vendor.",
"abstract": "Patient-access and documentation agents should be approved by owner and record path: PHI lifecycle, EHR permissions, recording consent, retention, clinician review and autonomous-use limits.",
"category": "AGENTS",
"sector": "Healthcare delivery",
"region": "USA",
"topics": [
"healthcare AI agents",
"patient access",
"clinical documentation",
"ambient scribing",
"EHR writeback",
"HIPAA",
"BAA",
"patient consent",
"recording consent",
"clinical review",
"payer administration",
"refill workflows",
"eligibility",
"intake",
"coding support",
"patient engagement",
"prior authorisation",
"consumer health data",
"TCPA",
"right of access",
"de-identification",
"liability"
],
"sources": [
{
"id": "S1",
"label": "Syllable - AI appointment scheduling",
"publisher": "Syllable",
"sourceClass": "vendor_docs",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof.",
"transformChain": [],
"url": "https://syllable.ai/use-cases/appointment-scheduling"
},
{
"id": "S2",
"label": "ActiumHealth / Syllable - MemorialCare multilingual AI case study",
"publisher": "ActiumHealth / Syllable",
"sourceClass": "customer_case_study",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "MemorialCare case supports multilingual patient-navigation scale and workflow expansion; vendor/customer story, not independent ROI proof.",
"transformChain": [],
"url": "https://www.actiumhealth.com/case-studies/memorialcare-multilingual-ai"
},
{
"id": "S3",
"label": "Hyro - healthcare AI agent platform",
"publisher": "Hyro",
"sourceClass": "vendor_docs",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof.",
"transformChain": [],
"url": "https://www.hyro.ai/healthcare/"
},
{
"id": "S4",
"label": "Hyro - Inova Health case study",
"publisher": "Hyro",
"sourceClass": "customer_case_study",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Inova case supports Epic/Cheers/NICE integration and appointment automation metrics; denominator and rework require buyer validation.",
"transformChain": [],
"url": "https://www.hyro.ai/case-study/inova-health/"
},
{
"id": "S6",
"label": "Infinitus - voice AI agents for healthcare",
"publisher": "Infinitus",
"sourceClass": "vendor_docs",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof.",
"transformChain": [],
"url": "https://www.infinitus.ai/product/ai-agents/"
},
{
"id": "S9",
"label": "Hippocratic AI - Polaris product page",
"publisher": "Hippocratic AI",
"sourceClass": "vendor_docs",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof.",
"transformChain": [],
"url": "https://www.hippocraticai.com/polaris/"
},
{
"id": "S12",
"label": "Assort Health - patient access platform",
"publisher": "Assort Health",
"sourceClass": "vendor_docs",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof.",
"transformChain": [],
"url": "https://www.assorthealth.com/platform"
},
{
"id": "S19",
"label": "S10.AI - CRUSH AI medical scribe",
"publisher": "S10.AI",
"sourceClass": "vendor_docs",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof.",
"transformChain": [],
"url": "https://s10.ai/solution/ai-medical-scribe"
},
{
"id": "S21",
"label": "S10.AI - pricing and ROI calculator",
"publisher": "S10.AI",
"sourceClass": "pricing_page",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "S10.AI page supplies public provider-month price signals; phone-agent, enterprise and integration economics remain quote-dependent.",
"transformChain": [],
"url": "https://s10.ai/pricing/roi-calculator"
},
{
"id": "S25",
"label": "Microsoft Learn - what is Microsoft Dragon Copilot",
"publisher": "Microsoft Learn",
"sourceClass": "vendor_docs",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof.",
"transformChain": [],
"url": "https://learn.microsoft.com/en-us/industry/healthcare/dragon-copilot/about/"
},
{
"id": "S27",
"label": "Abridge - generative AI for clinical conversations",
"publisher": "Abridge",
"sourceClass": "vendor_docs",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof.",
"transformChain": [],
"url": "https://www.abridge.com/"
},
{
"id": "S28",
"label": "Ambience Healthcare - Houston Methodist enterprise rollout",
"publisher": "Ambience Healthcare",
"sourceClass": "customer_case_study",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Houston Methodist/Ambience case supports enterprise ambient deployment metrics; vendor-published and methodology-sensitive.",
"transformChain": [],
"url": "https://www.ambiencehealthcare.com/blog/ambience-healthcare-announces-houston-methodist-enterprise-rollout-of-ai-platform-to-enhance-patient-care-and-clinical-workflows"
},
{
"id": "S30",
"label": "HHS OCR - business associates",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "HHS OCR business-associate guidance supports written-assurance, permitted-use and safeguard requirements for PHI vendors.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/for-professionals/privacy/guidance/business-associates/index.html"
},
{
"id": "S31",
"label": "HHS OCR - HIPAA Security Rule NPRM",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/for-professionals/security/hipaa-security-rule-nprm/index.html"
},
{
"id": "S32",
"label": "HHS OCR - 42 CFR Part 2 overview",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "HHS Part 2 page supports the February 16, 2026 compliance date and OCR complaint path for SUD-record handling.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/part-2/index.html"
},
{
"id": "S33",
"label": "ONC - HTI-1 final rule",
"publisher": "ONC",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://healthit.gov/regulations/hti-rules/hti-1-final-rule/"
},
{
"id": "S34",
"label": "ONC - decision support interventions test method",
"publisher": "ONC",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://healthit.gov/test-method/decision-support-interventions/"
},
{
"id": "S35",
"label": "FDA - January 2026 final clinical decision support software guidance",
"publisher": "FDA",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "FDA CDS guidance supports the device boundary for diagnosis, treatment, order and recommendation functions. Source row carries the January 2026 final-guidance status used for CDS boundary claims.",
"transformChain": [],
"url": "https://www.fda.gov/regulatory-information/search-fda-guidance-documents/clinical-decision-support-software"
},
{
"id": "S36",
"label": "American Medical Association - using health AI in the exam room",
"publisher": "American Medical Association",
"sourceClass": "professional_guidance",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Professional or governance guidance; use for adoption controls, risk-management posture and buyer diligence, not binding law unless stated.",
"transformChain": [],
"url": "https://www.ama-assn.org/practice-management/digital-health/using-health-ai-exam-room-what-doctors-should-consider"
},
{
"id": "S37",
"label": "Joint Commission and CHAI - responsible AI adoption guidance (published September 2025)",
"publisher": "Joint Commission and CHAI",
"sourceClass": "professional_guidance",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Professional or governance guidance; use for adoption controls, risk-management posture and buyer diligence, not binding law unless stated. Source path carries a September 2025 publication signal; checked date remains 2026-06-01.",
"transformChain": [],
"url": "https://www.jointcommission.org/en-us/knowledge-library/news/2025-09-jc-and-chai-release-initial-guidance-to-support-responsible-ai-adoption"
},
{
"id": "S38",
"label": "NIST - AI Risk Management Framework",
"publisher": "NIST",
"sourceClass": "policy_framework",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Professional or governance guidance; use for adoption controls, risk-management posture and buyer diligence, not binding law unless stated.",
"transformChain": [],
"url": "https://www.nist.gov/itl/ai-risk-management-framework"
},
{
"id": "S39",
"label": "FTC - complying with the Health Breach Notification Rule",
"publisher": "FTC",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://www.ftc.gov/business-guidance/resources/complying-ftcs-health-breach-notification-rule-0"
},
{
"id": "S45",
"label": "HHS OCR - HIPAA Privacy Rule summary",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/for-professionals/privacy/laws-regulations/index.html"
},
{
"id": "S46",
"label": "HHS OCR - HIPAA Security Rule summary",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/for-professionals/security/laws-regulations/index.html"
},
{
"id": "S47",
"label": "HHS OCR - HIPAA Breach Notification Rule",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/for-professionals/breach-notification/index.html"
},
{
"id": "S48",
"label": "HHS OCR - HIPAA de-identification guidance",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/for-professionals/special-topics/de-identification/index.html"
},
{
"id": "S49",
"label": "HHS OCR - HIPAA right of access and designated record set",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/for-professionals/faq/2045/does-an-individual-have-a-right-to-access/index.html"
},
{
"id": "S50",
"label": "CMS - interoperability and prior authorization final rule",
"publisher": "CMS",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "CMS-0057-F page supports 2026 prior-authorisation operational duties and 2027 API context for impacted payers.",
"transformChain": [],
"url": "https://www.cms.gov/priorities/key-initiatives/burden-reduction/policies-and-regulations/cms-interoperability-and-prior-authorization-final-rule-cms-0057-f"
},
{
"id": "S51",
"label": "FCC - AI-generated voices and TCPA declaratory ruling",
"publisher": "FCC",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "FCC declaratory ruling supports AI-generated voice treatment under TCPA artificial or prerecorded voice restrictions.",
"transformChain": [],
"url": "https://docs.fcc.gov/public/attachments/FCC-24-17A1_Rcd.pdf"
},
{
"id": "S52",
"label": "American Medical Association - 2026 physician AI survey",
"publisher": "American Medical Association",
"sourceClass": "professional_guidance",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "AMA 2026 physician AI survey supports adoption, privacy, validation, liability, training and involvement gates.",
"transformChain": [],
"url": "https://www.ama-assn.org/system/files/physician-ai-sentiment-report.pdf"
},
{
"id": "S53",
"label": "American Medical Association - physician concern over payer AI prior authorization",
"publisher": "American Medical Association",
"sourceClass": "professional_guidance",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Professional or governance guidance; use for adoption controls, risk-management posture and buyer diligence, not binding law unless stated.",
"transformChain": [],
"url": "https://www.ama-assn.org/press-center/ama-press-releases/physicians-concerned-ai-increases-prior-authorization-denials"
},
{
"id": "S54",
"label": "National Academy of Medicine - health care AI code of conduct",
"publisher": "National Academy of Medicine",
"sourceClass": "policy_framework",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Professional or governance guidance; use for adoption controls, risk-management posture and buyer diligence, not binding law unless stated.",
"transformChain": [],
"url": "https://nam.edu/our-work/programs/leadership-consortium/health-care-artificial-intelligence-code-of-conduct/"
},
{
"id": "S55",
"label": "Coalition for Health AI - best-practice guides and testing frameworks",
"publisher": "Coalition for Health AI",
"sourceClass": "professional_guidance",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Professional or governance guidance; use for adoption controls, risk-management posture and buyer diligence, not binding law unless stated.",
"transformChain": [],
"url": "https://www.chai.org/blog/coalition-for-health-ai-chai-releases-new-best-practice-guides-and-testing"
},
{
"id": "S56",
"label": "Washington State Legislature - My Health My Data Act",
"publisher": "Washington State Legislature",
"sourceClass": "state_law",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://app.leg.wa.gov/RCW/default.aspx?cite=19.373&full=true"
},
{
"id": "S57",
"label": "ONC - HTI rules index",
"publisher": "ONC",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://healthit.gov/regulations/hti-rules/"
},
{
"id": "S58",
"label": "CMS - interoperability standards and prior authorization for drugs proposed rule",
"publisher": "CMS",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "CMS-0062-P fact sheet supports the June 15, 2026 comment deadline and proposed drug prior-authorisation/FHIR changes.",
"transformChain": [],
"url": "https://www.cms.gov/newsroom/fact-sheets/2026-cms-interoperability-standards-prior-authorization-drugs-proposed-rule"
},
{
"id": "S59",
"label": "HHS OCR - HIPAA minimum necessary requirement",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/for-professionals/privacy/guidance/minimum-necessary-requirement/index.html"
},
{
"id": "S60",
"label": "HHS OCR - reproductive health privacy status",
"publisher": "HHS OCR",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://www.hhs.gov/hipaa/for-professionals/special-topics/reproductive-health/index.html"
},
{
"id": "S66",
"label": "ONC - HTI-4 final rule",
"publisher": "ONC",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "ONC HTI-4 page supports certification criteria for electronic prior authorisation, e-prescribing and real-time prescription benefit.",
"transformChain": [],
"url": "https://healthit.gov/regulations/hti-rules/hti-4-final-rule/"
},
{
"id": "S67",
"label": "ONC - HTI-2 final rule",
"publisher": "ONC",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://healthit.gov/regulations/hti-rules/hti-2-final-rule/"
},
{
"id": "S68",
"label": "ONC - HTI-3 final rule",
"publisher": "ONC",
"sourceClass": "regulator",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance.",
"transformChain": [],
"url": "https://healthit.gov/regulations/hti-rules/hti-3-final-rule/"
},
{
"id": "S69",
"label": "NEJM AI - ambient AI scribes randomized trial",
"publisher": "NEJM AI",
"sourceClass": "independent_paper",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "Randomized ambient-scribe trial supports measured time-in-note, survey and inaccuracy findings across 238 physicians and 14 specialties.",
"transformChain": [],
"url": "https://ai.nejm.org/doi/10.1056/AIoa2501000"
},
{
"id": "S70",
"label": "JMIR Medical Informatics - ambient listening note quality study",
"publisher": "JMIR Medical Informatics",
"sourceClass": "independent_paper",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "JMIR study supports note-quality error classes, serious-risk notes and edit-burden metrics across AI-created notes.",
"transformChain": [],
"url": "https://medinform.jmir.org/2026/1/e86474"
},
{
"id": "S71",
"label": "arXiv - PhysicianBench real-world EHR agent benchmark",
"publisher": "arXiv",
"sourceClass": "independent_benchmark",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "PhysicianBench supports EHR-agent strict-performance caveats across 21 specialties and tool-call-heavy tasks.",
"transformChain": [],
"url": "https://arxiv.org/abs/2605.02240"
},
{
"id": "S72",
"label": "arXiv - CHI-Bench healthcare workflow agent benchmark",
"publisher": "arXiv",
"sourceClass": "independent_benchmark",
"accessedAt": "2026-06-01",
"status": "verified",
"access": "sample",
"caveat": "CHI-Bench supports low completion rates for policy-dense prior-authorisation, utilisation-management and care-management agent workflows.",
"transformChain": [],
"url": "https://arxiv.org/abs/2605.16679"
}
],
"evidence": [
{
"id": "E1",
"sourceId": "S1",
"claimIds": [
"C1",
"C2"
],
"summary": "Extracted signal: Syllable positions healthcare scheduling agents around EHR, CRM, practice-management and telephony connections, with appointment booking, patient matching and system writeback as product claims. Report inference: Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof. Mapped claim IDs: C1, C2, C4, C8, C11.",
"status": "verified",
"confidence": 0.88,
"access": "sample"
},
{
"id": "E2",
"sourceId": "S2",
"claimIds": [
"C2"
],
"summary": "Extracted signal: The MemorialCare case describes Syllable handling and routing central patient-navigation calls, expanding into appointment, refill and password-reset workflows, and reporting roughly 150,000 monthly calls. Report inference: MemorialCare case supports multilingual patient-navigation scale and workflow expansion; vendor/customer story, not independent ROI proof. Mapped claim IDs: C2, C6, C8.",
"status": "verified",
"confidence": 0.86,
"access": "sample"
},
{
"id": "E3",
"sourceId": "S3",
"claimIds": [
"C1",
"C2",
"C5"
],
"summary": "Extracted signal: Hyro markets responsible healthcare AI agents for patient access, call-centre automation, FAQs, scheduling, prescription support, analytics and live-agent handoff, with HIPAA-oriented compliance claims. Report inference: Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof. Mapped claim IDs: C1, C2, C5, C8, C11.",
"status": "verified",
"confidence": 0.86,
"access": "sample"
},
{
"id": "E4",
"sourceId": "S4",
"claimIds": [
"C2"
],
"summary": "Extracted signal: Hyro's Inova case reports Epic, Cheers CRM and NICE CXone integration, 50% appointment-management automation, 338,000 automated calls per month and 4,272 monthly staff hours gained. Report inference: Inova case supports Epic/Cheers/NICE integration and appointment automation metrics; denominator and rework require buyer validation. Mapped claim IDs: C2, C6, C8, C9.",
"status": "verified",
"confidence": 0.84,
"access": "sample"
},
{
"id": "E6",
"sourceId": "S6",
"claimIds": [
"C1",
"C2"
],
"summary": "Extracted signal: Infinitus focuses on complex healthcare calls for patients, payers and providers, including benefit verification, prior authorisation follow-up, formulary exceptions, appeals, adherence and patient welcome calls. Report inference: Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof. Mapped claim IDs: C1, C2, C8.",
"status": "verified",
"confidence": 0.87,
"access": "sample"
},
{
"id": "E9",
"sourceId": "S9",
"claimIds": [
"C1",
"C3"
],
"summary": "Extracted signal: Hippocratic AI's Polaris page positions healthcare agents for non-diagnostic, non-prescribing workflows and publishes usage-price signals of Polaris Pro as low as $9 per hour and Polaris Flash as low as $5 per hour. Report inference: Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof. Mapped claim IDs: C1, C3, C7, C9.",
"status": "verified",
"confidence": 0.84,
"access": "sample"
},
{
"id": "E12",
"sourceId": "S12",
"claimIds": [
"C1",
"C2"
],
"summary": "Extracted signal: Assort Health describes a patient-access platform spanning inbound calls, outbound outreach, scheduling, referrals, payments, intake, medication refill tasking, waitlist management and EHR/PMS workflow sync. Report inference: Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof. Mapped claim IDs: C1, C2, C4, C8.",
"status": "verified",
"confidence": 0.86,
"access": "sample"
},
{
"id": "E19",
"sourceId": "S19",
"claimIds": [
"C1",
"C3"
],
"summary": "Extracted signal: S10.AI's medical-scribe product supports encounter capture, structured clinical notes, coding support and EHR-ready documentation, while public material states clinicians review, edit and sign off. Report inference: Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof. Mapped claim IDs: C1, C3, C4, C7.",
"status": "verified",
"confidence": 0.83,
"access": "sample"
},
{
"id": "E21",
"sourceId": "S21",
"claimIds": [],
"summary": "Extracted signal: S10.AI publishes entry price points starting at $99 per provider per month for Basic and $150 for Pro with EHR integration, while enterprise and phone-agent economics require quote review. Report inference: S10.AI page supplies public provider-month price signals; phone-agent, enterprise and integration economics remain quote-dependent. Mapped claim IDs: C9.",
"status": "verified",
"confidence": 0.82,
"access": "sample"
},
{
"id": "E25",
"sourceId": "S25",
"claimIds": [
"C1",
"C3"
],
"summary": "Extracted signal: Microsoft Learn describes Dragon Copilot as creating draft clinical output from recordings for clinician review, requiring patient consent before recording and supporting embedded EHR workflows where users edit summaries inside the EHR. Report inference: Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof. Mapped claim IDs: C1, C3, C4, C7, C10.",
"status": "verified",
"confidence": 0.9,
"access": "sample"
},
{
"id": "E27",
"sourceId": "S27",
"claimIds": [
"C1",
"C3"
],
"summary": "Extracted signal: Abridge positions its ambient clinical-conversation product around multiple care settings, specialties, languages, integrated workflows and direct Epic use from Haiku to Hyperdrive. Report inference: Vendor product claim; use for workflow scope, integrations and shortlist relevance, not independent ROI or safety proof. Mapped claim IDs: C1, C3, C4, C7.",
"status": "verified",
"confidence": 0.84,
"access": "sample"
},
{
"id": "E28",
"sourceId": "S28",
"claimIds": [
"C1",
"C3"
],
"summary": "Extracted signal: Ambience reports Houston Methodist enterprise deployment across ambulatory, emergency and inpatient settings, with 80% of patient visits using Ambience, 40% lower documentation time, 33% lower after-hours documentation and 1.3 voluntary visits per clinician per day. Report inference: Houston Methodist/Ambience case supports enterprise ambient deployment metrics; vendor-published and methodology-sensitive. Mapped claim IDs: C1, C3, C6, C7, C9.",
"status": "verified",
"confidence": 0.84,
"access": "sample"
},
{
"id": "E30",
"sourceId": "S30",
"claimIds": [
"C5",
"C13"
],
"summary": "Extracted signal: HHS OCR explains that covered entities may disclose PHI to business associates only with satisfactory written assurances that PHI will be used for the engaged purpose and safeguarded from misuse. Report inference: HHS OCR business-associate guidance supports written-assurance, permitted-use and safeguard requirements for PHI vendors. Mapped claim IDs: C5, C11, C12, C13.",
"status": "verified",
"confidence": 0.94,
"access": "sample"
},
{
"id": "E31",
"sourceId": "S31",
"claimIds": [
"C5"
],
"summary": "Extracted signal: HHS OCR's HIPAA Security Rule NPRM says large breach reports increased by 102% from 2018 to 2023 and affected individuals increased by 1002%, framing cybersecurity and ePHI protection as patient-safety issues. Report inference: Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance. Mapped claim IDs: C5, C11, C12.",
"status": "verified",
"confidence": 0.92,
"access": "sample"
},
{
"id": "E36",
"sourceId": "S36",
"claimIds": [
"C3",
"C5"
],
"summary": "Extracted signal: The AMA advises that when AI tools are used during an encounter, especially ambient scribing, physicians should explain the technology, secure consent and reassure patients about HIPAA compliance. Report inference: Professional or governance guidance; use for adoption controls, risk-management posture and buyer diligence, not binding law unless stated. Mapped claim IDs: C3, C5, C11.",
"status": "verified",
"confidence": 0.88,
"access": "sample"
},
{
"id": "E45",
"sourceId": "S45",
"claimIds": [
"C5",
"C13"
],
"summary": "Extracted signal: HHS OCR's Privacy Rule summary defines PHI broadly across electronic, paper and oral forms, limits uses and disclosures by covered entities and business associates, and treats individually identifiable health and payment information as protected when linked to identifiers. Report inference: Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance. Mapped claim IDs: C5, C11, C12, C13, C14.",
"status": "verified",
"confidence": 0.93,
"access": "sample"
},
{
"id": "E46",
"sourceId": "S46",
"claimIds": [
"C5",
"C13"
],
"summary": "Extracted signal: HHS OCR's Security Rule summary requires regulated entities and business associates to protect ePHI with administrative, physical and technical safeguards, including access control, audit controls, authentication, transmission security, risk analysis and incident procedures. Report inference: Primary rule or regulator source; use for dates, legal duties and approval boundaries, not vendor performance. Mapped claim IDs: C5, C11, C12, C13, C14.",
"status": "verified",
"confidence": 0.93,
"access": "sample"
},
{
"id": "E50",
"sourceId": "S50",
"claimIds": [
"C2"
],
"summary": "Extracted signal: CMS' interoperability and prior authorisation final rule sets 2026 operational requirements and primarily 2027 API requirements for impacted payers, including APIs for patient, provider, payer-to-payer and prior-authorisation data exchange. Report inference: CMS-0057-F page supports 2026 prior-authorisation operational duties and 2027 API context for impacted payers. Mapped claim IDs: C2, C8, C10, C16.",
"status": "verified",
"confidence": 0.91,
"access": "sample"
},
{
"id": "E52",
"sourceId": "S52",
"claimIds": [
"C3",
"C5",
"C13"
],
"summary": "Extracted signal: AMA's 2026 physician AI survey reports more than 80% physician awareness or use of AI in practice, with documentation and summarisation prominent, while privacy, safety validation, liability clarity, training and physician involvement remain adoption gates. Report inference: AMA 2026 physician AI survey supports adoption, privacy, validation, liability, training and involvement gates. Mapped claim IDs: C3, C5, C12, C13, C15.",
"status": "verified",
"confidence": 0.88,
"access": "sample"
},
{
"id": "E58",
"sourceId": "S58",
"claimIds": [],
"summary": "Extracted signal: CMS' April 2026 proposed rule would extend many prior-authorisation requirements to drugs, propose FHIR standards for prior-authorisation transactions, require API endpoint reporting and set a public-comment deadline that closed on June 15, 2026. Report inference: CMS-0062-P fact sheet supports the June 15, 2026 comment deadline and proposed drug prior-authorisation/FHIR changes. Mapped claim IDs: C10, C16.",
"status": "monitor",
"confidence": 0.88,
"access": "sample"
},
{
"id": "E69",
"sourceId": "S69",
"claimIds": [
"C3"
],
"summary": "Extracted signal: A pragmatic randomized trial of 238 outpatient physicians across 14 specialties found Nabla reduced time-in-note by 9.5% versus control, DAX showed no significant time-in-note change versus control, both scribes improved some secondary survey measures, and occasional clinically significant inaccuracies required vigilance. Report inference: Randomized ambient-scribe trial supports measured time-in-note, survey and inaccuracy findings across 238 physicians and 14 specialties. Mapped claim IDs: C3, C7, C9, C15.",
"status": "verified",
"confidence": 0.9,
"access": "sample"
},
{
"id": "E70",
"sourceId": "S70",
"claimIds": [
"C3"
],
"summary": "Extracted signal: A JMIR Medical Informatics prospective pilot across 7,545 AI-created notes found 94.7% of evaluated notes free from significant errors, but accidental omissions, hallucinations and serious-risk errors still appeared; median AI-generated words changed was 9.0%, with wide physician variation and 14.9% of measured notes unedited. Report inference: JMIR study supports note-quality error classes, serious-risk notes and edit-burden metrics across AI-created notes. Mapped claim IDs: C3, C7, C9, C15.",
"status": "verified",
"confidence": 0.88,
"access": "sample"
},
{
"id": "E71",
"sourceId": "S71",
"claimIds": [
"C3"
],
"summary": "Extracted signal: PhysicianBench tests EHR-grounded LLM agents across 21 specialties and clinical workflow types with an average of 27 tool calls per task; the best reported pass@1 was 46%, while open-source models reached at most 19%. Report inference: PhysicianBench supports EHR-agent strict-performance caveats across 21 specialties and tool-call-heavy tasks. Mapped claim IDs: C3, C7, C15, C17.",
"status": "verified",
"confidence": 0.84,
"access": "sample"
},
{
"id": "E72",
"sourceId": "S72",
"claimIds": [],
"summary": "Extracted signal: CHI-Bench tests long-horizon healthcare workflows in provider prior authorisation, payer utilisation management and care management across 20 simulated healthcare apps and 87 MCP tools; the best agent resolved 28.0% of tasks, no agent cleared 20% on strict pass^3 and single-session execution fell to 3.8%. Report inference: CHI-Bench supports low completion rates for policy-dense prior-authorisation, utilisation-management and care-management agent workflows. Mapped claim IDs: C10, C16, C17.",
"status": "verified",
"confidence": 0.84,
"access": "sample"
}
],
"claims": [
{
"id": "C1",
"statement": "Healthcare AI agents should be routed as four related markets: patient access, payer and provider administration, clinical documentation, and clinical-adjacent engagement.",
"confidence": 0.9,
"evidenceIds": [
"E1",
"E3",
"E6",
"E9",
"E12",
"E19",
"E25",
"E27",
"E28"
],
"access": "sample"
},
{
"id": "C2",
"statement": "Patient-access agents can be useful for scheduling, routing, refills, eligibility, intake, reminders and outreach, but the approval gate is identity, PHI handling, EHR writeback, escalation and auditability.",
"confidence": 0.9,
"evidenceIds": [
"E1",
"E2",
"E3",
"E4",
"E6",
"E12"
],
"access": "sample"
},
{
"id": "C3",
"statement": "Ambient clinical documentation should remain a clinician-reviewed draft-record workflow unless the buyer has separate evidence for clinical decision support, orders or autonomous recommendations.",
"confidence": 0.9,
"evidenceIds": [
"E9",
"E19",
"E25",
"E27",
"E28",
"E36"
],
"access": "sample"
},
{
"id": "C5",
"statement": "HIPAA badges are not workflow approval: BAA scope, subprocessors, model training, data retention, transcript and audio storage, call-recording consent, Part 2 and breach-notification posture must be tied to the exact workflow.",
"confidence": 0.9,
"evidenceIds": [
"E30",
"E31",
"E36",
"E45",
"E46"
],
"access": "sample"
},
{
"id": "C13",
"statement": "Organisation approval should be segmented by owner: workflow ambition, clinical boundary, EHR write authority, PHI flow, security, economics, vendor evidence and operating metrics belong in different approval files.",
"confidence": 0.88,
"evidenceIds": [
"E30",
"E45",
"E46",
"E52"
],
"access": "sample"
}
],
"colophon": {
"checkedAt": "2026-06-01",
"latestRevisionAt": "2026-06-17T00:00:00Z",
"reviewStandard": "expert_human_reviewed",
"kinetics": {
"speed": "fast",
"impact": 0.91,
"momentum": 0.89,
"businessImpact": "Healthcare-agent buying has shifted from vendor fit to record-path approval: PHI flow, EHR write authority, clinician review and accepted-output evidence now decide production.",
"rateOfChange": "CMS prior-authorisation rulemaking [E50, E58], ONC certification movement [E66], FDA CDS guidance [E35], ambient-scribe studies [E69, E70] and healthcare-agent benchmarks [E71, E72] can move approval before annual budget cycles.",
"from": "chatbot, scribe and call-centre pilots",
"toward": "approved record paths with named owners, PHI contracts, EHR permissions, consent, clinician review and accepted-output measurement"
},
"watch": [],
"gapAnalysis": "Independent longitudinal healthcare-agent evidence remains thin for PHI lifecycle controls, EHR writeback reliability, clinician edit burden, benchmark pass rates in real EHR workflows, prior-authorisation denial logic, escalation safety, incident rates and cost per accepted workflow or accepted note."
},
"body": {
"sections": [
{
"id": "executive-summary",
"title": "Decision summary",
"access": "sample",
"dek": "Record path comes before vendor fit.",
"claimIds": [
"C1",
"C2",
"C3",
"C5",
"C13"
],
"evidenceIds": [
"E1",
"E25",
"E30",
"E50",
"E69",
"E72"
],
"blocks": [
{
"id": "executive-summary-narrative",
"evidenceIds": [
"E1",
"E2",
"E4",
"E6",
"E19",
"E25",
"E27",
"E28",
"E36",
"E50",
"E52",
"E58",
"E69",
"E70",
"E71",
"E72"
],
"type": "section_narrative",
"lead": "The first action is to approve the record path.",
"body": [
"Approve healthcare agents as workflow authority: answer, transfer, book, create a task, draft a note, suggest a code, support a prior authorisation or recommend care. Each step changes the owner, evidence file and stop sign.",
"Healthcare AI agents divide into four useful lanes: patient access, payer and provider administration, clinical documentation, and clinical-adjacent engagement. Patient access has the closest fit with customer-experience automation, while ambient documentation touches the encounter and EHR write path. [E1, E25]",
"The patient-access proof is named workflow deployment: scheduling, reminders, refills, intake, eligibility, benefit verification, referral follow-up, dental booking and member outreach. [E4, E13] Ambient documentation needs a different proof file: recording consent, capture quality, clinician edits, coding support and final sign-off. [E25, E28]",
"Prior authorisation is becoming a live governance lane, and recent public evidence has hardened the benchmark signal. CMS dates and physician concern about payer AI widen the file; note-quality studies and healthcare-agent benchmarks make assistive approval the default for clinical action. [E50, E53] [E69, E72]"
],
"detail": [
"The approval forum should approve expansion by owner: operations for access flow, clinical leadership for care boundaries, EHR ownership for write permissions, privacy and security for PHI lifecycle, and finance for accepted-output economics."
]
},
{
"id": "executive-action-table",
"title": "Approval starts with owners",
"evidenceIds": [
"E25",
"E30",
"E31",
"E45",
"E46",
"E52"
],
"type": "table",
"columns": [
{
"key": "decision",
"label": "Decision to make",
"align": "left"
},
{
"key": "owner",
"label": "Accountable owner",
"align": "left"
},
{
"key": "evidence",
"label": "Evidence file",
"align": "left"
},
{
"key": "stopSign",
"label": "Stop sign",
"align": "left"
}
],
"rows": [
{
"decision": "Approve workflow ambition",
"owner": "Operating owner with patient-access or service-line leadership",
"evidence": "Allowed workflows, expansion limits, patient-trust posture and staffing impact.",
"stopSign": "The pilot can expand before the workflow owner can explain what the agent may do."
},
{
"decision": "Approve clinical boundary",
"owner": "Clinical leadership, nursing leadership and speciality leads",
"evidence": "Allowed scripts, blocked clinical topics, clinician review, escalation and incident classes.",
"stopSign": "The agent can influence diagnosis, treatment, orders, triage or coding without a separate clinical approval."
},
{
"decision": "Approve record write authority",
"owner": "Record-system owner and informatics",
"evidence": "Read/write matrix, sandbox tests, rollback, failed-write reconciliation and audit exports.",
"stopSign": "The vendor shows an EHR logo but not field-level writeback evidence."
},
{
"decision": "Approve PHI lifecycle",
"owner": "Privacy, security and legal",
"evidence": "BAA, minimum-necessary policy, consent, retention, subprocessors, de-identification and breach path.",
"stopSign": "Audio, transcripts, prompts or analytics exports sit outside the approved PHI flow."
},
{
"decision": "Approve economics",
"owner": "Finance, revenue-cycle and operations analytics",
"evidence": "Cost per accepted workflow or signed note after repeat contact, edits, rework and retained staff.",
"stopSign": "The business case counts containment, booked visits or draft notes before acceptance."
}
]
},
{
"id": "executive-record-authority-bridge",
"evidenceIds": [
"E25",
"E30",
"E36"
],
"type": "paragraph",
"text": "Owner approval only names who signs. The next decision is what each work product is allowed to change.",
"variant": "small"
},
{
"id": "executive-decision-table",
"title": "Record authority changes approval",
"evidenceIds": [
"E1",
"E25",
"E30",
"E36"
],
"type": "table",
"columns": [
{
"key": "decision",
"label": "Decision",
"align": "left"
},
{
"key": "whatItMeans",
"label": "What it means",
"align": "left"
},
{
"key": "buyerMove",
"label": "Buyer move",
"align": "left"
}
],
"rows": [
{
"decision": "Answer and route",
"whatItMeans": "The agent explains approved content, identifies intent and sends the patient to the right queue.",
"buyerMove": "Approve with content grounding, escalation logs and call-recording rules."
},
{
"decision": "Complete an administrative transaction",
"whatItMeans": "The agent books, cancels, reschedules, creates a task, captures intake or starts a refill workflow.",
"buyerMove": "Require identity checks, EHR/PMS writeback tests, rollback, exception queues and BAA scope."
},
{
"decision": "Draft the clinical record",
"whatItMeans": "The system records or receives the encounter and produces a note, summary, coding cue or structured field.",
"buyerMove": "Require patient consent, clinician review, edit tracking, retention controls and final-signature ownership."
},
{
"decision": "Recommend clinical action",
"whatItMeans": "The system influences orders, diagnoses, treatment, triage or clinical judgement.",
"buyerMove": "Treat as a separate clinical-decision-support approval with FDA, ONC, governance and safety evidence."
}
]
},
{
"id": "executive-demo-traps-bridge",
"evidenceIds": [
"E4",
"E25",
"E28",
"E69",
"E70"
],
"type": "paragraph",
"text": "Demo metrics are useful only after those paths are clear. The remaining executive test is whether sales claims survive accepted-output proof.",
"variant": "small"
},
{
"id": "executive-traps-table",
"title": "The demo hides the chart risk",
"evidenceIds": [
"E25",
"E30",
"E31",
"E36"
],
"type": "table",
"columns": [
{
"key": "trap",
"label": "Trap",
"align": "left"
},
{
"key": "whatItHides",
"label": "What it hides",
"align": "left"
},
{
"key": "betterTest",
"label": "Better test",
"align": "left"
}
],
"rows": [
{
"trap": "HIPAA-compliant banner",
"whatItHides": "BAA scope, subprocessors, model-provider terms, audio retention and transcript access.",
"betterTest": "Trace one live PHI flow from capture to deletion and inspect contract language."
},
{
"trap": "EHR integration logo",
"whatItHides": "Read-only access, custom services work, failed writebacks, duplicate bookings and audit-export limits.",
"betterTest": "Force each target action in a sandbox and reconcile records, logs and exception queues."
},
{
"trap": "High automation rate",
"whatItHides": "False resolution, repeat calls, abandoned transfers, staff rework and excluded edge cases.",
"betterTest": "Measure accepted workflows after follow-up windows and human-review cost."
},
{
"trap": "Ambient note quality",
"whatItHides": "Consent practice, noisy-room capture, omitted negatives, copy-forward risk and clinician edit burden.",
"betterTest": "Compare draft, edited and signed notes with a clinician-owned rubric."
}
]
},
{
"id": "route-definition-callout",
"title": "Route definition",
"evidenceIds": [
"E1",
"E25",
"E30",
"E50",
"E69",
"E72"
],
"type": "callout",
"tone": "neutral",
"body": "Route means the approved path from work to accepted output: the source set, data boundary, model or vendor channel, tools, reviewer, evidence record and reapproval trigger that make an AI result usable."
},
{
"id": "executive-summary-section-callout",
"title": "Approve the write path",
"evidenceIds": [
"E25",
"E30",
"E36"
],
"type": "callout",
"tone": "warning",
"body": "Patient conversation fluency does not grant chart authority. The chart path, review owner and consent model decide the route."
}
]
},
{
"id": "decision-ownership",
"title": "Decision ownership",
"access": "sample",
"dek": "Approve healthcare agents by record authority and accountable owner.",
"claimIds": [
"C5",
"C13"
],
"evidenceIds": [
"E25",
"E30",
"E31",
"E45",
"E46",
"E52"
],
"blocks": [
{
"id": "decision-ownership-narrative",
"evidenceIds": [
"E25",
"E30",
"E31",
"E45",
"E46",
"E52"
],
"type": "section_narrative",
"lead": "The live decision is which record path the organisation is willing to automate.",
"body": [
"Healthcare agents cross operating boundaries quickly. A scheduling agent can become a chart-writing system when it books into the EHR. An ambient scribe can become a revenue-cycle tool when it suggests codes. A member-engagement agent can become a clinical-risk system when it handles symptoms, medication questions or care-management prompts.",
"Approval should therefore sit with named owners. Operations owns access flow and staffing impact. Clinical leadership owns documentation, care boundaries and clinician review. The EHR owner owns read, write, rollback and audit permissions. Privacy, security and legal own PHI flow, retention, consent, subprocessors and incident response. Finance owns the accepted-output denominator, not the vendor's headline automation metric.",
"Workflow leaders should build the evidence file by lane. The approval forum should approve, block or route each workflow before procurement turns a pilot into production."
]
},
{
"id": "decision-ownership-table",
"title": "Every approval file needs an owner",
"evidenceIds": [
"E25",
"E30",
"E31",
"E45",
"E46",
"E52"
],
"type": "table",
"columns": [
{
"key": "owner",
"label": "Owner",
"align": "left"
},
{
"key": "approvalRight",
"label": "Approval right",
"align": "left"
},
{
"key": "evidenceRequired",
"label": "Evidence required",
"align": "left"
},
{
"key": "relevantSections",
"label": "Where to work",
"align": "left"
}
],
"rows": [
{
"owner": "Enterprise operating owner",
"approvalRight": "Workflow ambition, rollout sequence and patient-trust posture.",
"evidenceRequired": "Allowed workflows, blocked uses, service-level targets, staffing plan and expansion rule.",
"relevantSections": "Decision frame, patient-access workflows and operating model."
},
{
"owner": "Clinical leadership",
"approvalRight": "Clinical boundary, documentation policy, care escalation and clinician accountability.",
"evidenceRequired": "Consent model, clinician review, speciality rubric, incident classes and final-signature rule.",
"relevantSections": "Clinical documentation, human acceptance and evaluation controls."
},
{
"owner": "EHR and informatics owner",
"approvalRight": "Read, task, scheduling, note, coding and order permissions.",
"evidenceRequired": "Writeback tests, duplicate handling, rollback, audit logs and right-of-access classification.",
"relevantSections": "Record write path and regulatory controls."
},
{
"owner": "Privacy, security and legal",
"approvalRight": "PHI lifecycle, BAA, consent, Part 2, TCPA, state privacy and breach posture.",
"evidenceRequired": "Minimum-necessary map, subprocessors, retention, de-identification, opt-out and incident procedure.",
"relevantSections": "Regulatory and contract controls, known unknowns and source ledger."
},
{
"owner": "Finance and revenue-cycle",
"approvalRight": "Economic case, prior-authorisation impact and accepted-output denominator.",
"evidenceRequired": "Cost per accepted workflow or note, denial-reason capture, appeal evidence and revenue attribution method.",
"relevantSections": "Pricing and economics, patient-access workflows and recommended next steps."
},
{
"owner": "Workflow lead",
"approvalRight": "Sandbox design, staff handoff, QA sampling and operating metrics.",
"evidenceRequired": "Test scripts, queue design, exception reporting, escalation samples and weekly production dashboard.",
"relevantSections": "Evaluation controls, operating model and vendor directory."
}
]
}
]
},
{
"id": "what-changed",
"title": "What changed",
"access": "sample",
"dek": "The market has moved from front-door automation to record-path control.",
"claimIds": [
"C1",
"C13"
],
"evidenceIds": [
"E28",
"E50",
"E52",
"E58"
],
"blocks": [
{
"id": "what-changed-narrative",
"evidenceIds": [
"E28",
"E50",
"E52",
"E58"
],
"type": "section_narrative",
"lead": "Patient access, ambient documentation and prior authorisation are starting to meet in the same workflow file.",
"body": [
"Ambient documentation has moved from novelty to enterprise deployment evidence. Newer public signals from Rush/Suki [E62], Ochsner/DeepScribe [E64], Houston Methodist/Ambience [E28] and M Health Fairview/Nabla [E29] expand the category beyond isolated scribe pilots, though most evidence remains vendor-published and uneven on edit burden, speciality mix and control groups.",
"The payer/provider administration lane is getting more regulated and more technical. CMS has 2026 prior-authorisation operational duties and mostly 2027 API requirements for impacted payers, and CMS-0062-P would extend many requirements to drugs if finalised. [E50, E58]",
"EHR and documentation vendors are moving toward broader control planes. Oracle Health is bundling clinical and administrative agent functions, while Abridge and Highmark/AHN are linking ambient documentation with real-time prior authorisation. [E61, E63]",
"Human acceptance is no longer a soft issue. Physician AI use is rising, but privacy, validation, liability, training and clinician involvement still decide whether a technically capable system can survive production. [E52, E55]"
]
},
{
"id": "what-changed-table",
"title": "The approval file is getting wider",
"evidenceIds": [
"E50",
"E52",
"E58"
],
"type": "table",
"columns": [
{
"key": "change",
"label": "Change",
"align": "left"
},
{
"key": "whyItMatters",
"label": "Why it matters",
"align": "left"
},
{
"key": "buyerMove",
"label": "Buyer move",
"align": "left"
}
],
"rows": [
{
"change": "Prior-authorisation APIs and metrics",
"whyItMatters": "CMS prior-authorisation APIs and metrics mean agents can affect access to care and denial evidence, not just administrative speed. [E50, E58]",
"buyerMove": "Require source criteria, denial-reason capture, appeal evidence and human-review logs."
},
{
"change": "AI voice treated as artificial or prerecorded voice",
"whyItMatters": "Outbound patient calls and outreach need consent, identification and opt-out controls under the FCC's AI voice TCPA ruling. [E51]",
"buyerMove": "Tie consent source, suppression lists and emergency exceptions to each campaign."
},
{
"change": "EHR incumbents bundling agents",
"whyItMatters": "The shortlist may become an EHR control-plane decision rather than a standalone vendor search. [E61, E63]",
"buyerMove": "Test write authority, audit exports and rollback before accepting platform convenience."
},
{
"change": "Physician adoption rising",
"whyItMatters": "The barrier is less awareness than trust, liability and clinical workflow fit. [E52, E55]",
"buyerMove": "Include training, feedback, post-market monitoring and clinician ownership in procurement."
}
]
},
{
"id": "what-changed-section-callout",
"title": "Prior authorisation joins ambient notes",
"evidenceIds": [
"E50",
"E58"
],
"type": "callout",
"tone": "info",
"body": "When documentation feeds authorisation, the proof file must cover both the clinical note and the payer decision that follows."
}
]
},
{
"id": "decision-frame",
"title": "Decision frame",
"access": "sample",
"dek": "Choose the lane before choosing the vendor.",
"claimIds": [
"C1",
"C2",
"C3"
],
"evidenceIds": [
"E1",
"E6",
"E19",
"E25",
"E28",
"E30",
"E36"
],
"blocks": [
{
"id": "decision-frame-narrative",
"evidenceIds": [
"E1",
"E6",
"E19",
"E25",
"E30",
"E36"
],
"type": "section_narrative",
"lead": "Healthcare agents should be sorted by the work product they leave behind.",
"body": [
"An access agent leaves a booked appointment, routed call, task, refill request, intake packet or eligibility note. A payer-administration agent leaves a benefit-verification result, prior-authorisation status or member-engagement record. An ambient scribe leaves a draft clinical record. A clinical-adjacent agent may leave education, outreach or follow-up tasks that start to resemble care management.",
"Output type sets the proof packet. As the result moves into the medical record or clinical decision loop, the buyer needs consent, review, source attributes, risk management, incident records and regulatory boundary-setting. [E25, E35]",
"A generic customer-service agent is the wrong default for PHI work. A healthcare vendor still needs diligence, but it starts with concrete controls: BAA, EHR permissions, consent, escalation, retention and clinical owner."
]
},
{
"id": "healthcare-agent-routing-tree",
"title": "Record writes split booking from notes",
"caption": "Scheduling writes need identity, availability, referral, insurance, duplicate, rollback and BAA proof; clinical-note writes need consent, retention, clinician review and final-signature proof.",
"evidenceIds": [
"E1",
"E4",
"E6",
"E19",
"E25",
"E30",
"E31",
"E36"
],
"type": "decision_tree",
"lead": "Classify the data artefact and write authority before vendor fit.",
"steps": [
{
"prompt": "Does the workflow touch PHI, patient identifiers, audio, transcripts or EHR data?",
"tone": "info",
"branches": [
{
"label": "Yes",
"outcome": "Open the BAA, retention, subprocessor, recording-notice and consent file before lane approval.",
"next": "Q2",
"tone": "warning"
},
{
"label": "No",
"outcome": "Use generic CX controls only if there is no PHI, no recording and no EHR access.",
"next": "Q2",
"tone": "info"
}
]
},
{
"prompt": "Does the output recommend diagnosis, treatment, triage, orders, coding or care-plan action?",
"tone": "warning",
"branches": [
{
"label": "Yes",
"outcome": "Route to clinical-decision or revenue-cycle approval with source attributes, risk management, clinician override and CDS/FDA boundary review.",
"tone": "blocked"
},
{
"label": "No",
"outcome": "Continue to documentation and transaction checks.",
"next": "Q3",
"tone": "info"
}
]
},
{
"prompt": "Does the agent create a draft clinical note, encounter summary or structured clinical field?",
"tone": "blocked",
"branches": [
{
"label": "Yes",
"outcome": "Use the clinical-documentation lane: patient consent, audio/transcript retention, draft-to-signed diff, clinician edit tracking and final-signature ownership.",
"tone": "warning"
},
{
"label": "No",
"outcome": "Continue to patient-access write checks.",
"next": "Q4",
"tone": "info"
}
]
},
{
"prompt": "Can the agent book, reschedule, cancel, waitlist or create/close an administrative EHR/PMS task?",
"tone": "warning",
"branches": [
{
"label": "Yes",
"outcome": "Use the patient-access transaction lane: identity proofing, provider rules, referral/insurance checks, duplicate prevention, rollback, audit logs and exception queues.",
"tone": "warning"
},
{
"label": "No",
"outcome": "Continue to answer-and-route checks.",
"next": "Q5",
"tone": "info"
}
]
},
{
"prompt": "Is the agent only answering approved content, collecting intake or routing to staff?",
"tone": "info",
"branches": [
{
"label": "Yes",
"outcome": "Use patient-access assist controls: grounding, no-answer behaviour, escalation, call notice, transcript sampling and BAA review where PHI appears.",
"tone": "success"
},
{
"label": "No",
"outcome": "Hold in read-only assist until the owner, output, data flow and write authority are defined.",
"tone": "monitor"
}
]
}
]
},
{
"id": "cx-boundary-bridge",
"evidenceIds": [
"E1",
"E3",
"E30",
"E45"
],
"type": "paragraph",
"text": "Healthcare record-path approval ends where generic contact-centre economics begin. Broader CCaaS vendor structure and voice-agent workflow controls belong in the adjacent CX report.",
"variant": "small"
},
{
"id": "related-reports",
"evidenceIds": [],
"type": "report_link",
"reportId": "rpt_ai_customer_experience_workflows_2026_05",
"slug": "ai-customer-service-workflows",
"label": "AI for customer-service workflows",
"dek": "Use the broader CX report for CCaaS, helpdesk, outcome pricing and generic voice-agent market structure.",
"relation": "Related report"
}
]
},
{
"id": "record-write-path",
"title": "Record write path",
"access": "locked",
"dek": "The signature proof is field-level permission.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "record-write-path-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "patient-access-workflows",
"title": "Patient-access workflows",
"access": "locked",
"dek": "The strongest public proof is administrative, not diagnostic.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "patient-access-workflows-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "clinical-documentation",
"title": "Clinical documentation",
"access": "locked",
"dek": "Ambient AI should make the chart easier to finish, not easier to trust blindly.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "clinical-documentation-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "benchmark-private-tests",
"title": "Benchmark and private tests",
"access": "locked",
"dek": "Public scores are now useful because they argue against autonomous approval.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "benchmark-private-tests-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "human-acceptance-liability",
"title": "Human acceptance and liability",
"access": "locked",
"dek": "Adoption is rising, but trust still has to be earned workflow by workflow.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "human-acceptance-liability-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "vendor-lanes",
"title": "Vendor lanes",
"access": "locked",
"dek": "The shortlist should separate access, administration, documentation and engagement.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "vendor-lanes-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "regulatory-contract-controls",
"title": "Regulatory and contract controls",
"access": "locked",
"dek": "Healthcare approval is a contract and workflow file.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "regulatory-contract-controls-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "evaluation-controls",
"title": "Evaluation and controls",
"access": "locked",
"dek": "The private test should include the worst safe version of the workflow.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "evaluation-controls-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "pricing-economics",
"title": "Pricing and economics",
"access": "locked",
"dek": "The useful unit is accepted workflow or accepted note.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "pricing-economics-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "operating-model",
"title": "Operating model",
"access": "locked",
"dek": "The governance owner should be named before the pilot.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "operating-model-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "known-unknowns",
"title": "Known unknowns",
"access": "locked",
"dek": "Evidence gaps should become approval gates.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "known-unknowns-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "recommended-next-steps",
"title": "Recommended next steps",
"access": "locked",
"dek": "A safe pilot is narrow, measured and record-aware.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "recommended-next-steps-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "vendor-directory",
"title": "Vendor directory",
"access": "locked",
"dek": "Vendor relevance starts after lane, record path and test owner are named.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "vendor-directory-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "source-ledger",
"title": "Source ledger",
"access": "locked",
"dek": "Use source class and caveat to separate rules, studies, benchmarks and vendor claims.",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "source-ledger-locked",
"title": "Locked section",
"evidenceIds": [],
"type": "callout",
"tone": "locked",
"body": "Purchase or subscribe to unlock this section."
}
]
},
{
"id": "glossary",
"title": "Glossary",
"access": "sample",
"claimIds": [],
"evidenceIds": [],
"blocks": [
{
"id": "glossary-table",
"title": "Terms used in this report",
"caption": "Definitions are reader aids, not separate evidence claims.",
"evidenceIds": [],
"type": "table",
"columns": [
{
"key": "term",
"label": "Term",
"align": "left"
},
{
"key": "meaning",
"label": "Plain meaning",
"align": "left"
}
],
"rows": [
{
"term": "Route",
"meaning": "The approved path from work to accepted output: source set, data boundary, model or vendor channel, tools, reviewer, evidence record and reapproval trigger."
},
{
"term": "API",
"meaning": "Application programming interface: a governed software connection used by an app or workflow to call a model or service."
},
{
"term": "OCR",
"meaning": "Optical character recognition: converting scanned or image text into machine-readable text."
}
]
}
]
}
]
}
},
"gap_analysis": "Independent longitudinal healthcare-agent evidence remains thin for PHI lifecycle controls, EHR writeback reliability, clinician edit burden, benchmark pass rates in real EHR workflows, prior-authorisation denial logic, escalation safety, incident rates and cost per accepted workflow or accepted note.",
"kinetics": {
"speed": "fast",
"impact": 0.91,
"momentum": 0.89,
"businessImpact": "Healthcare-agent buying has shifted from vendor fit to record-path approval: PHI flow, EHR write authority, clinician review and accepted-output evidence now decide production.",
"rateOfChange": "CMS prior-authorisation rulemaking [E50, E58], ONC certification movement [E66], FDA CDS guidance [E35], ambient-scribe studies [E69, E70] and healthcare-agent benchmarks [E71, E72] can move approval before annual budget cycles.",
"from": "chatbot, scribe and call-centre pilots",
"toward": "approved record paths with named owners, PHI contracts, EHR permissions, consent, clinician review and accepted-output measurement"
},
"watch_signals": [],
"sample_findings": [
"Healthcare AI agents should be approved by owner and record path: answer, task, schedule, draft note, coding cue, prior-authorisation support or clinical recommendation."
]
}
```