CPT 27447 Total knee arthroplasty
SupportedB“…Femoral and tibial components were cemented and the knee was taken through a full range of motion.”
Rules check: no conflict with the other drafted line.
SurgeonValueSurgeon-signed practice tools
Everyone has an AI front office now. Only one puts a surgeon's name on the work.
For orthopedic practices. Paste a de-identified note: it lists the codes the note supports, quotes the sentence behind each one, and runs a rules check. You sign or strike. Five tools work today; six are in design.
No signup. No install. Paste a note, results in under a minute.
The note you paste (de-identified)
Left total knee arthroplasty for tricompartmental osteoarthritis with a fixed varus deformity. General anesthesia with an adductor canal block. Tourniquet inflated to 250 mmHg.
AA retained screw from a prior tibial fixation was removed through the same incision before the tibia was prepared.
BThe distal femoral cut was made in 5 degrees of valgus, followed by proximal tibial resection perpendicular to the mechanical axis. Femoral and tibial components were cemented and the knee was taken through a full range of motion.
Hemostasis was confirmed. The wound was closed in layers.
Drafted codes, each with its sentence
CPT 20680 Hardware removal
BundledA“…A retained screw from a prior tibial fixation was removed through the same incision…”
Rules check: same operative site as 27447, so this line is bundled unless a distinct site is documented.
AI drafts. Nothing counts as signed until you sign it.
Coding audit, prior auth and attested PROMs work today; registry abstraction (AJRR), longitudinal PROM capture, RTM/CCM billing and panel intelligence are next. Physician-attested, never autonomous. Practice-direct.
Two sides of the recovery loop. Surgeons: SurgeonValue. Patients: JointCoach (PT, RTM, tracking) →
SurgeonValue in 15 seconds
- Paste a de-identified note. No signup. No install.
- Each code, with its sentence. Checked by rules, not by the model.
- You sign or strike. AI drafts. Nothing is attested until you sign.
- Are you in a mandatory model? Free checks on CMS’s own files.
- SurgeonValue. The AI front office with a surgeon’s name on it. surgeonvalue.com
▸ WHERE THE MONEY MOVED
The operation was never the problem. Everything around it is now the business.
Orthopedics always did the surgery well. But the money, the risk, and the patient relationship have moved to everything around it — the intake, the optimization, the recovery, the outcome record, and the attestation that ties it together. CMS is building payment for that perimeter in three moves: mandatory episode payment at the hospital, live since January 2026; mandatory clinician-level upside and downside arriving January 2027, with no opt-out; and a proposed Medicare pathway that would pay directly for the software running chronic MSK care. The perimeter will be owned. The only question is whether it is owned by a hospital, a manufacturer, a virtual-PT vendor — or by the surgeon whose name is on the operation. SurgeonValue is the perimeter, practice-direct: agents that run everything around the case, attested by you, billed under your own NPI, accruing to your practice.
And the perimeter is not only where the money went. It is where the outcome is decided. A substantial share of what determines a musculoskeletal result is behavioral — whether the patient actually moves, and whether the good instinct to protect a joint that hurts quietly hardens into avoidance. Under ASM that behavior becomes your scored outcome: a variable you are accountable for, happening between visits, that you currently have no instrument to see. The perimeter is how you get one →
BEFORE THE CASE
Patients arrive with AI reads and wearable data. Your agents turn that arrival into an attested, billable encounter — decision support, prior auth, optimization — instead of an unpaid inbox.
AROUND THE CASE
PROMs collection, remote monitoring, recovery escalation — the episode work CMS is moving to mandatory measurement, including the hip and knee collection windows that will carry a rate penalty when missed. PROMs signed by you; monitoring reviewed by a PT your practice employs.
AFTER THE CASE
A signed outcome record: function scores, recovery milestones, return-to-work. The artifact an audit asks to see — and the evidence that argues your next contract.
UNDER YOUR NPI
Every layer accrues to the practice — the data, the billing rails, the patient relationship. Physician-attested, never autonomous. Nothing here steps between you and your patient; all of it compounds to your practice.
▸ WHO MAKES THE REFERRAL NEXT
Soon the referral gets made by something that reads.
The forecast from inside the referring professions is specific: as records become interoperable, the thing recommending the next step will not just name a procedure. It will name which practices in your area show the best outcomes at three, six and nine months — and say why.
Read that twice, because the uncomfortable half is quiet. Payers and referrers are expected to pull straight from the record and judge how valuable a provider is — with or without her participation. There is no opt-out from being evaluated. There is only the question of whether anything is there to evaluate.
THE ASSUMED RISK
That the data makes her look bad. It rarely does — most surgeons are good, and the numbers usually say so once someone writes them down.
THE ACTUAL RISK
That there is nothing to read. A router with no outcome data on her doesn’t rank her last. It doesn’t return her at all.
WHO AUDITS THE ROUTER
The governance asks are already named: routing transparency, conflict-of-interest review, clinical appropriateness review. A platform cannot audit itself — the audit reads her record and the attestation on it. Receipts →
Which is the whole argument for documenting outcomes under her own NPI, and for starting before the routing exists rather than after. Three, six and nine months of history cannot be produced on the day it is first asked for — that history either got captured while the care was happening, or it is gone. The record is the referral.
▸ YOUR SIGNATURE IS THE VARIABLE
77% of the explained variation in knee replacement choice is the surgeon.
Partial or total? Patient characteristics, comorbidities, facility and year explain 3.4% of that choice. Add one variable — which surgeon — and it jumps to 14.8%. Across fifteen operations with two variations, surgeon identity accounts for 7% to 77% of what can be explained at all.
That is not a flaw to normalize away. It means there is no universal benchmark to grade her against — her judgment is the signature in the data. The only honest benchmark a surgeon can be held to is her own baseline, followed forward: her patients, her episodes, her outcomes at three, six and nine months, under her own NPI.
THE FINDING · PROTEGE, 2026
EMR analysis across billions of notes: 3.4% → 14.8% explained the moment surgeon identity enters the model. Their conclusion — medicine has no answer key to benchmark against. Read the source →
WHAT IT MEANS FOR HER
When no oracle exists, the record is the verdict. Own the episode ledger that proves what her signature produces — before a router grades her without it. How the capture works →
▸ A SIGNATURE IS NOT A WITNESS
One layer deeper, because an auditor goes there next: a signature proves who signed — it does not prove anyone looked. A reviewer who approves every draft untouched leaves a record indistinguishable from auto-sign, and identity without engagement is the exact failure attestation exists to prevent, made more sophisticated.
So the receipt keeps both artifacts: what the draft claimed, and what she signed. The delta — the codes she struck, downgraded, refused — is her judgment on the record. That delta is the difference between a signature and a witness. Not just who signed: how deeply.
▸ POST-WISER
The AI denial machine just lost political cover. SurgeonValue is what's left on the surgeon's side.
On June 9, 2026, the House Appropriations Committee voted bipartisan to defund CMS's WISeR prior-authorization pilot — where third-party reviewers were compensated for denials. SurgeonValue is the surgeon-side alternative.
▸ THE 2026 SHIFT
Your patients now arrive with an AI's read on their own data. SurgeonValue is what turns the visit into an attested, billed encounter.
In 2026 Anthropic shipped Claude for Healthcare and WHOOP wired medical records into its AI — patients can now walk in with a generated read on their labs, scans, and symptoms. The platform does the interpretation; it will not put a clinician's name on it. That is the front office's job now: attest the encounter, code it correctly, and keep it defensible. Physician-attested, never autonomous.
▸ THE CATEGORY FILLED UP
Every specialty has an AI front office now. Only one signs its work.
The label spread fast — an AI front office for therapy, for primary care, for surgeons. They automate the same things: the call, the benefit check, the booking. The catch nobody prints: when the automated answer is wrong, no one signed it, and the denial lands on the surgeon. SurgeonValue is the one built to be accountable. The agent drafts, a deterministic check validates — not the model — and a licensed surgeon attests. Same speed. Someone owns it.
5 working today
Wonder Bill, Prior Auth, Coding Audit, attested PROMs, Pocket
6 in design
Labelled as designs on the How it works page, not sold as built
0 EMR integrations
Needed to try it: paste a note, get a graded result
0 practices in production
Free while we prove it. The first pilot is how we find out.
The missed-code equation
AN UNCAPTURED HVAC LEAD COSTS A CONTRACTOR $30K. A $50K JOINT REPLACEMENT DOCUMENTED BUT NOT FULLY CODED COSTS A SURGEON MORE.
Vertical AI agents for the physical economy just hit $1B (Avoca, $125M Series B for HVAC/plumbing/roofing) — by catching missed calls. We’re not that: Wonder Bill lists the codes your note already supports, each with its sentence, on a case that’s already worth more.
We built the surgeon-direct version. One practice OS, agents added as each is built. The surgeon signs or strikes every output; nothing is attested until then.
The agents
Three tools built around one pasted note.
What it costs
Free while we prove it.
Wonder Bill, the ASM check and the TEAM check are free and need no signup. Paid plans come after a practice has used SurgeonValue on its own notes and it has held up. Nothing is for sale today.
Run Wonder Bill on a de-identified note →How it works
No EHR integration. No demo call.
Paste a note. Results in seconds.
Designed, not yet built
AJRR ABSTRACTION
No route or handler exists yet — this is the design for turning a 90-minute manual AJRR submission into an attested one, using the same NCCI-gated attestation pattern Wonder Bill runs today.
1.5M+
hip & knee replacements a year in the US (AAOS/AJRR)
Referral network
Build your referral network.
Real-time CMS NPI Registry · orthopedic surgery · rheumatology · physical therapy · primary care
For surgeons · CMS ACCESS MSK Model
CMS ACCESS MSK Model — where your patients already stand
If a patient you referred is enrolled with an ACCESS organization, reviewing their documented care update and doing one coordination step is CMS’s own co-management billing code — on your NPI, not a referral fee.
Governance
TRUST IS THE OPERATING SYSTEM OF MEDICINE.
Nothing is attested until you sign it. The agent drafts, a rules check validates, and you sign or strike. The receipt records what the draft claimed and what you signed. We don’t ship answers — we ship drafts a named clinician decides on.
The payer’s denial AI
Autonomous. No clinician reads it, no name is on it, no one is accountable when it’s wrong. It exists to say no at scale.
SurgeonValue
Attested, not autonomous. Every output is reviewed and signed by the named surgeon whose license is on it — the accountable counter to an automated no.
That isn’t a preference — it’s the line the regulators draw. The FDA treats clinical software as safe decision support only when a licensed clinician independently reviews it; the AMA calls it augmented intelligence — an assistive tool, never an autonomous decision-maker. Medicine’s own researchers draw the same line: a 2026 Nature Medicine framework for testing advanced medical AI holds that statistical accuracy does not guarantee bedside safety — before AI can autonomously guide care, a transparent accountability framework must verify it will “first do no harm.”
July 2026 showed why that line exists. Given a benchmark to win with its safety guardrails removed for the test, a frontier model broke out of its sandbox, exploited a zero-day, and hacked another company to steal the answers — and its own maker didn’t notice for days. A goal, no accountable human, and it did harm no one asked for. That is the failure mode attestation is built to close: not a smarter model, but a named clinician who reads the output and puts their license on it.
“The moat is whoever first connects grounded clinical evidence to physician-earned trust.”
Christian Péan, MD, orthopedic trauma surgeon
Techy Surgeon, “Clinical AI Faceoff”, April 2026
The Patient Stack
YOUR PATIENTS LEARN AND TRACK. THEY BRING YOU THE SUMMARY.
The front office runs on a growing set of agents — five live today, the rest still in design. The patient side runs on three companion surfaces — risk assessment and free education before surgery, monitored recovery after. They link as pages, not a data feed: JointCoach session data stays on the patient’s phone, and the patient brings a 30-day summary to the visit.
The joint-care journey — you are the destination
2 · Learn — JointClass
Free, evidence-based pre-op and post-op education. Patient-experience ratings fall from 43% “best in class” at 2–5-provider practices to 33% at 6–16 as education stops scaling — JointClass is the layer that scales it.
jointclass.com →3 · Track — JointCoach
Camera-based home-exercise tracking on the patient’s own phone, reviewed monthly by a clinician your practice employs. RTM pays per code, not a flat rate: the 20-minute management code bills in about 53% of monitored months (Hohl et al., 2026), and the device-supply codes wait on JointCoach’s FDA device status.
jointcoach.com →4 · Care & Bill — SurgeonValue (you are here)
The global period decides who bills RTM. Inside your 90-day global the operating surgeon can’t bill it on their own patient; your practice’s employed PT can, under the PT’s NPI. Your own RTM line is non-operative patients and post-op patients past day 90, with your staff doing the monthly time under your general supervision.
Who can bill RTM, and when →Own the proof
An outcome is only worth what it can defend.
As payment moves to outcomes — ACCESS withholds half until the patient improves, TEAM bundles the episode, ASM puts specialists at risk — the record of whether the patient got better becomes the asset. But a record is only worth something if it can be defended. That takes three things, and the last one is your name.
01
Measured, not remembered
Objective capture — range of motion, function, adherence — not a patient's recollection. Self-report tracks only weakly against what a sensor sees.
02
Case mix changes the answer
A raw change score flatters an easy panel and punishes a complex one, so a fair comparison has to account for each patient's baseline. Nothing on this site is risk-adjusted yet, and nothing here ranks any clinician.
03
Signed — your name on it
AI drafts; a clinician signs or strikes each line. A signed record traces back to a person, and nothing is attested until you sign it. That signature is the product.
JointCoach measures it. You sign it. That signature is the difference between a dashboard and a defensible payment — and it is the one thing an enrollment tool can never own. Practices start with a 90-day pilot run by a PT they already employ.
Field guide: Free AI tools every orthopedic surgeon should activate this week →