LifeX Ventures · Research tool
Edit the assumptions directly. The model recalculates workforce mix, technically addressable task capacity, realized capacity, and the primary-care day instantly. Baseline: an illustrative 100,000-person integrated health provider.
A size-adjusted estimate of revenue, payroll, care-delivery labor COGS, and the annual value of capacity shifted to AI. Revenue and margin are scenario estimates—not forecasts for the named comparison organization. Released capacity must actually be converted into cost avoidance to create the modeled profit impact.
| Illustrative operating P&L bridge | Annual estimate |
|---|---|
| Estimated revenue | — |
| Baseline operating profit | — |
| Add: realizable labor value | — |
| Less: AI operating cost | — |
| Illustrative post-AI operating profit | — |
| Illustrative post-AI operating margin | — |
| Workforce COGS bridge | Annual estimate |
|---|---|
| Baseline labor in COGS | — |
| Less: realizable COGS labor value | — |
| Add: AI operating cost | — |
| Illustrative post-AI labor COGS | — |
The $300,000 default revenue-per-employee assumption is a rounded provider-group scenario anchor; integrated payer-provider systems can run higher. The 3% baseline operating margin is editable. Wage anchors are rounded national BLS role benchmarks. AI cost excludes integration, change management, validation, and vendor minimums.
Headcount by job family, scaled automatically to the selected workforce size. Evidence W1–W6 ↓
Point headcount × midpoint of the addressable task range.
Low assumes narrow use and weak automation. Guarded is an early-deployment case. Base uses the central research estimate. High assumes broad use, strong integration, and favorable workflows.
Concentration slider0% spreads the same average exposure evenly across everyone. 25–50% creates some specialization. 55–75% clusters much of the work. 80–100% puts nearly all affected work into specialist roles.
Healthcare examplePrior authorization may be concentrated among dedicated specialists, while documentation is spread across nearly every nurse. The same total hours imply very different cloneable-role counts.
Other-industry exampleInvoice entry can sit with dedicated accounts-payable clerks, while email is spread across every office worker.
Task exposure controls the mean share of work affected. Concentration controls whether that exposure is spread across everyone or clustered into specialist roles. The model preserves the same mean exposure while changing the number of people above the displacement threshold.
| Job family | Headcount | Task exposure | Concentration | Technical FTE-eq. | Augmented capacity | Cloneable roles |
|---|
Choose task-exposure levels below; the chart shows the resulting releasable share of a measured 480-minute day. Evidence T1–T9 ↓
The model now separates four concepts that should not be collapsed.
Concentration is the crucial bridge. Even exposure produces slack across many people; concentrated exposure creates a smaller number of potentially redesignable roles.
Time savings only become headcount when they are concentrated in the same roles, contiguous enough to schedule, and large enough that the residual work can be reassigned. Otherwise they become capacity, service, or slack.
ATMs automated routine cash handling, but lower branch operating costs encouraged more branches and tellers shifted toward relationship and sales work. James Bessen's analysis is a classic example of task automation expanding output and changing jobs rather than translating one-for-one into headcount.
Bessen, “Toil and Technology” · Public-domain image
AI can shorten interpretation or drafting time while radiologists remain responsible for review and rising imaging demand absorbs capacity. In a Swedish mammography study, AI-supported screening reduced human workload by 44% while retaining radiologist oversight—large task savings without a simple 44% headcount conclusion.
Evidence summary and study context · NIH public-domain imageA time-motion study recorded 5,061 tasks in 58.7 hours; 44.7% of time was spent on clinical-information-system work. The workload was highly interrupted and fragmented, so removing clicks or seconds does not automatically create a schedulable block or eliminate the need for the clinician.
Clinical workflow fragmentation studyOECD research emphasizes the task content of individual jobs rather than occupation averages. Two people with the same title can have very different exposure—exactly the distinction controlled by the concentration slider.
OECD Employment OutlookImage credits: Library of Congress / Gottscho-Schleisner collection and National Cancer Institute / Bill Branson. Both are identified by their repositories as public domain in the United States.
Large technology, services, outsourcing, and virtual-care companies that can reduce work or offer an alternative delivery path—plus relevant companies from the LifeX portfolio. Select any company to visit its site.
One click selects a bundled low, guarded, base, or high estimate for each activity.
| Activity | Minutes/day | Task exposure | Gross releasable | Automation / augmentation |
|---|
The model triangulates organizational disclosures, federal occupational data, clinician time studies, workflow research and task-level AI frameworks. No single source contains the complete answer. Each source below states what it supports—and what it does not.
Observed anchorSupports the 100,000+ denominator and integrated-provider footprint: 28 hospitals and 1,000+ outpatient facilities. It does not disclose a complete job-family census.
Northwell Health, About UsBoundary checkReports a 20,000+ physician network. We do not treat this as 20,000 employed physicians; the model distinguishes network affiliation from employee headcount.
Northwell, Physician CareersObserved anchorBLS industry staffing counts by occupation underpin the relative scale of nurses, physicians, support, laboratory, pharmacy, facilities and administrative work in hospitals.
BLS OEWS, Hospitals, May 2023ContextProvides occupation definitions and the demand backdrop; BLS projects roughly 1.9 million annual healthcare openings over 2024–34.
BLS Occupational Outlook, HealthcareContextSupports the constraint that released capacity will coexist with rising demand: healthcare support and practitioner occupations are among the fastest-growing groups.
BLS Employment Projections, 2024–34ContextDocuments HRSA's workforce simulation approach and nursing supply/demand assumptions. Used as a planning context, not as Northwell headcount.
HRSA Nursing Model ComponentsThe slider displays the closest reported employee count—not a claim that the organizations have the same workforce mix, operating model, or scope. Counts are rounded and reflect each system's own terminology and reporting period.
Scale markerOfficial overview reports 50,000 employees.
NewYork-Presbyterian, About UsScale markerOfficial leadership profile reports 81,000+ caregivers worldwide.
Cleveland Clinic leadershipScale markerOfficial careers overview reports 100,000+ team members.
Northwell Health, About UsScale markerYear-end 2025 disclosure reports more than 120,000 caregivers.
Providence 2025 disclosureScale markerOfficial reporting describes approximately 134,000 associates.
Ascension overviewScale markerOfficial system profile reports approximately 150,000 employees.
CommonSpirit system profileScale markerOfficial 2026 reporting describes more than 240,000 employees.
Kaiser Permanente overviewNYC practice markerThe Astoria- and Bayside-based dental group is listed in the 11–50 employee range.
Precision Dental NYC profileNYC practice markerThe East Harlem cancer-treatment partnership is listed in the 51–200 employee range.
New York Proton Center profileNYC practice markerThe pediatric group reports more than 50 neighborhood offices and is listed in the 201–500 employee range.
Tribeca PediatricsNYC practice markerThe Midtown-founded specialty group reports 120+ locations and 400+ providers; its company profile places employment in the 1,001–5,000 range.
Schweiger Dermatology GroupNYC hospital markerThe hospital's official careers overview reports more than 5,000 employees.
NYU Langone Hospital—BrooklynNYC system markerThe Brooklyn health system reports more than 7,000 employees across over 80 locations.
Maimonides Health overviewNYC practice markerA real single-location Manhattan dental practice used only as a small-practice scale analogy; no exact employee count is asserted.
SoHo Dental GroupNational scale markerA named national healthcare enterprise replaces the former generic largest-scale placeholder. Its mixed payer, care-delivery, pharmacy, and technology model is not directly comparable to a provider-only group.
UnitedHealth Group annual reportsMeasured workflowDirect observation across ambulatory practices found substantial desk/EHR work and established the widely cited relationship between face time and EHR/desk work.
Annals of Internal Medicine, 2016Measured workflowEvent logs for 142 family physicians, validated with observation: 355 EHR minutes per weekday, including documentation/order/coding and inbox work. Central calibration for the PCP day.
Annals of Family Medicine, 2017Measured workflowObserved 20 physicians in five clinics before and after EHR implementation; informs visit-level time allocation and cautions against assuming digitization automatically reduces total time.
AHRQ summary and study referenceMeasured workflowObserved 23 PCPs and 211 visits; found EHR work occupied 35% of in-person, 46% of phone and 39% of video visit time.
Journal of General Internal MedicineMethodDevelops a shared physician/nurse time-motion taxonomy for direct care, documentation and other work. Supports consistent task categories across clinical roles.
AMIA Proceedings, 2019Measured workflowStudied 767 nurses across 36 medical-surgical units; identifies documentation, medication administration and care coordination as major nursing-time pools.
The Permanente Journal, 2011Measured workflowContinuous observation of nursing activities highlights documentation location, task switching and interruptions—important constraints on simplistic automation assumptions.
BMC Health Services Research, 2019Intervention evidencePre/post time-motion study found a mobile workflow reduced EHR documentation by 4.1 minutes per observed hour, illustrating plausible task-level—not job-level—capacity release.
JMIR Medical Informatics, 2021Administrative burdenCMS cites estimates of roughly 13 hours per week and 700 hours annually per provider associated with prior authorization, supporting a high addressable range for forms and revenue-cycle workflows.
CMS Electronic Prior AuthorizationAI calibrationFrames expected change as task shifting and augmentation more than full job automation, and highlights trust, skills, validation and implementation constraints.
OECD, Artificial Intelligence and the Health WorkforceAI calibrationAnalyzes 55.5 million job postings and classifies health occupations by augmentation/automation susceptibility. Supports higher augmentation than automation ratings for clinicians.
OECD AI Paper No. 36, 2025Task ontologyProvides occupation-specific task statements, ratings and work activities. It supports decomposing jobs into activities rather than assigning one automation score to an entire occupation.
U.S. Department of Labor, O*NET DatabaseExternal benchmarkHealthcare analysis estimates 35% technical automation potential but a lower midpoint adoption outcome, reinforcing the model's separation of technical potential from realization.
Transforming Healthcare with AI, 2020External benchmarkMGI-style activity analysis finds healthcare automation potential around the mid-30% range and emphasizes that many occupations are transformed at the task level rather than eliminated.
Automation at Scale, 2019Use-case mapDocuments candidate workflows across clinical operations, analytics, IT, procurement, talent, finance and consumer service; supports the relative ranking of information-heavy work.
McKinsey, Generative AI in Healthcare, 2023GovernanceRequires human autonomy, safety, transparency, accountability, inclusiveness and sustainability. Supports low job-level automation ratings for accountable clinical work.
WHO Guidance, 2021Evidence synthesisCatalogues measures and interventions for documentation burden and illustrates variation by specialty, setting, data source and metric definition.
AHRQ Technical Brief No. 47Wage anchorNational and hospital-industry occupational wage estimates inform rounded hourly wage assumptions for each modeled job family.
BLS OEWS, May 2024Cost loadBLS reports wages at approximately 70% of private-industry employer compensation, implying a roughly 43% load on wages for benefits and payroll costs.
BLS Employer Costs, March 2026Revenue anchorUPMC reports $34 billion of 2025 operating revenue and 100,000 employees, implying about $340,000 per employee for its integrated provider and insurance system. The model uses a lower, rounded $300,000 default and makes it adjustable.
UPMC Facts & Figures, 2025
