Model your return
Your workflow.
Your numbers. Every assumption visible.
Compare the work your team does today with the work that remains with KAIS. Include gathering records, finding the correct reviewer, checking policy, preparing forms, reviewing evidence and submitting. A prepared packet can remove many steps. Missing evidence and required clinical decisions still take time.
Your practice and costs
The $50 hourly default is an illustration: $6,500 monthly employer cost divided by 130 productive task hours. BLS reports $46.89 per hour worked in private-industry total compensation. Replace it with your payroll and active-task hours; each workflow can use a different rate.
Prices are before tax. List onboarding is $1,500 per provider with no minimum. COMMAND PLUS is $3,500 per provider per month with no monthly minimum. Groups of 10 or more use the group rate; past 16 the figure is an estimate and the written quote is final. A year paid in advance takes a further 3 percent off; this page shows list. List prices may change; your signed order form holds for its term.
Your authorization team
The 100-hour default is the group model's own figure per provider at the full benchmark gain: eligibility 26.67, remittance 20, claim preflight 13.33, phone calls 12, faxes 10.67, chart review 8, claim rework 8, medical-necessity appeals 2.13 (about 100.8 hours), from the same CAQH 2024 minutes. Enter 0 to see authorization alone.
Starting figures are published benchmarks: company size from Kareo's 2023 billing-industry report (277 companies), request volume from the AMA's 2025 survey (40 per physician per week), minutes from CAQH 2024 (24 manual, 16 partially electronic, 10 fully electronic, the KAIS target). The subscription is priced on the providers you bill for, above.
Work performed each month
Count active minutes across everyone involved, once per task. Include record retrieval, AI preparation, verification, rework and submission. Exclude passive waiting. Use the supported share to keep unsupported cases at today's workload. A missing required time holds the result.
For monthly totals, use arithmetic mean minutes from comparable cases, including slow cases and rework. Keep the median and slow-case times as separate context. The hourly rate values the roles whose time is released; this worksheet does not calculate staffing changes between roles. Use actual avoided costs for a cash claim.
Build the case time, step by step
Use these logs to replace a whole-case guess. Enter 0 explicitly for a step that does not apply. A blank physician-letter or signature step is not treated as completed work. The same person may fill several roles; count their minutes once.
Authorization: plan, policy, route, packet and submission
Medical and pharmacy requests may have different reviewers, forms and destinations. Include routing work in today's time. Use a KAIS time only for a route actually supported and verified in your installation; this worksheet does not establish PBM criteria, vendor access or automatic submission. Required provider review remains.
Medical-necessity appeal: records, physician letter, AI and submission
Practitioner-reported example: approximately 16 minutes with general-purpose AI for an experienced appeals representative. This is not a national average or a measured KAIS result. Confirm which steps that timing includes, especially the physician letter, sign-off and submission; add missing active work below. A payer's requirement must be checked for the case.
A shorter wait may matter for care and turnaround. It does not become staff hours saved or additional revenue automatically.
Revenue authorization protects
Each example is a published Medicare national amount: CMS 2026 Physician Fee Schedule for 72148 (5.74 RVUs) and 27447 (34.71 total RVUs; $1,159 in AAHKS's table of 2026 values) at the $33.4009 conversion factor, and the CMS Part B ASP file for October 2026 (denosumab $29.856 per mg). The surgeon fee excludes the facility payment, which is larger. The figure is applied to every request above, so use the average across your mix, not your highest-value service. The default is the three examples averaged with equal weight, $1,047.48, because practices that carry authorization work mix imaging, procedures and drugs; the lumbar MRI alone is the conservative case. Pick the one closest to your authorizations, or enter your own average; an oncology or infusion practice should enter its own drug mix, because a single infusion can be $10,000 or more.
Requests come from your authorization rows above (or, for a billing company, your monthly authorizations). Defaults: KFF 2024, 7.7 percent of Medicare Advantage prior-authorization requests denied, 11.5 percent of denials appealed and 80.7 percent of appeals overturned, so about 90.7 percent of a denial is never recovered; a blended payment of $1,047.48, the average of the three published Medicare examples above (lumbar MRI $191.72, total knee surgeon fee $1,159.35, denosumab 60 mg $1,791.36). A biologic or a surgery puts thousands at stake per request; enter your own average. If you appeal every denial, the share never recovered is closer to 19.3 percent (the appeals KFF found upheld). The prevented share is a target: measure it on your own denials. Half the gains halves it.
Cash is a separate calculation
Use evidenced avoided hiring, overtime or outsourcing costs. Do not add the capacity value again. Redeploying salaried staff alone does not reduce payroll. Do not also claim the same released hours as both avoided staffing and new treatment capacity.
Benchmark: Change Healthcare's 2017 Denials Index found 9 percent of hospital claims initially denied, 63 percent of denials recoverable, and about $118 spent per claim on appeals. Use your own denial report for the inputs.
Cases × net received × unrecoverable share, less additional delivery costs. Do not use billed charges. A denial you would recover later is not a permanently lost collection. Subtract costs of additional care or drugs when incurred; do not subtract existing sunk costs twice. Outcomes and causation require measurement. The workload what-if buttons do not increase these cash inputs.
Whose work changes
KAIS takes the lookups, assembly, forms, faxes and rework. People keep the judgment, the signatures and the patient.
The first 12 months
Adoption builds over the first months, and follow-up savings (rework, appeals and status calls) arrive a month later, after the first requests go out. Set your own ramp.
Compare an upgrade on the difference
Compare the same entered workload with the next plan. No extra volume or faster time is invented. Complete timings for the added workflow before drawing a conclusion. Any separate chart-review quote is counted only when selected on COMMAND.
No comparison selected.
"We could build this with AI ourselves."
A general model writes a letter. It does not know which of 683,997 authorization rule records applies to this member, which of 129 payers or which review vendor decides it, or what changed in the policy last quarter. Building that means engineers (BLS puts the median software developer at $135,980 a year before benefits), a payer-policy library to collect and keep current, model and fine-tuning spend, testing, security review and a BAA. And every one of those costs keeps running after launch.
KAIS ships it as the package: the rule library, the vendor routing, the Reviewer Sheet and the request your team sends, maintained for you, per provider per month.
Assumptions & sources
Make the comparison hold up
- Use your existing tools as the baseline. If staff already use AI, portals or an EHR interface, time that workflow. Do not compare it with an artificially manual process.
- Time consecutive cases, not just easy successes. Separate administrative appeals, medical-necessity appeals, medical services, medical-benefit drugs and pharmacy requests. Include incomplete cases and corrections.
- Use the same start and finish. Count every participant's active time from obtaining the needed inputs to the agreed handoff or confirmed submission. Record the scope. A drafted letter and a submitted packet are different endpoints.
- Hold quality constant. An authorized reviewer checks the correct patient, plan, benefit, service, destination, policy version, evidence, signatures and packet. A fast incorrect result is not a saving.
- Measure the exceptions. Track supported share, missing-record cases, routing holds and rework. Record sample size, median and slow-case times. Small demo samples are preliminary, not proof of long-term ROI or payer outcomes.
- Reconcile to actual costs. Verify the signed quote, platform fees, payroll, avoided hires and collections. A seven-day demo can time workflow steps; longer follow-up is needed for denial, payment and retention outcomes.
Source register and boundaries
Pricing basis: current list prices and the group schedule. List pricing is a model, not a binding quote. Model version . No public benchmark here establishes KAIS's processing time, clinical accuracy, approval rate or financial outcome.