04 / Test public value
Make a pilot case you can challenge.
Start with a real workflow, compare the full staff effort before and after assistance, and test whether the capacity returned justifies the operating and implementation costs. The numbers below are illustrative until your team measures them.
Workflow under review
Choose a use case
Select a workflow from the explorer or pick one here.
Modeled as an API implementation.
Editable assumptions
Describe one repeatable unit of work
Use the same unit for baseline and assisted time. Assisted time must include drafting, checking, corrections, escalation, and sign-off.
Start here: replace the example volume, staff time, and costs with estimates for your workflow. The starting values are identical for every use case.
Illustrative planning estimate — verify every assumption before use
Three-year modeled net capacity value
Capacity value proxy after modeled costs, not cash savings
Time difference × monthly volume × 12 × realization factor = annual capacity hours. Hours × loaded hourly cost = annual capacity value proxy. Three-year net = three years of that value minus three years of recurring cost and one-time implementation.
Pilot evidence
Measure quality alongside time
The use case’s suggested measure and validation questions travel here. Set a baseline, define an acceptable quality threshold, and record what staff had to correct before treating any modeled value as evidence.
Choose a representative, bounded workflow with cleared material and a human reviewer.
Staff time including review; completeness; corrections; service outcome.
Data permissions, source quality, accessibility, human approval, and agency policy.
Record what the pilot finds
Use aggregate measures and a human reviewer’s conclusion. These notes do not change the value calculation.
This browser saves the scenario locally. Do not enter protected case details.
Time returned is capacity for higher-value work, better service, or backlog reduction. It is not a payroll reduction or an appropriated-dollar saving. “Capacity ROI” divides modeled three-year net capacity value by modeled three-year cost. A positive ratio does not establish that service quality, equity, safety, or eligibility controls are acceptable. The example inputs are invented; replace them with observed baseline and pilot data. Calculations run in your browser; the API cost handoff places the estimate in the URL. Before scaling, record the agency owner, approved use-case inventory entry if applicable, data classification, impact assessment, reviewer sign-off, and monitoring plan under the rules that apply to that agency.
Measurement approach informed by GAO’s AI accountability framework ↗ and NIST’s AI RMF Measure guidance ↗. Neither endorses this estimate.