Healthcare AI companies need NPI-verified physicians to attest clinical documents. The draft is already written. Your job is to catch what the AI got wrong. You review under your own malpractice coverage — no entity policy is in force yet, and the queue holds any case you are not covered for. No shifts. No minimum commitment.
Verify Your NPI to StartYou review under your own coverage • Pilot: synthetic cases, not open for live cases yet
Claude AI drafts LMNs, RTM reviews, CCM summaries, and care plans from structured patient data.
The original patient data and the AI interpretation are shown side by side. You see exactly what the AI used and what it concluded.
Approve (right), reject (left), or edit (up) — and that completes the task. Approve signs it: your determination becomes the anchored record. Edit means a fixable gap: you document what needs to change — you don't rewrite the document yourself — your note travels with the case, and it comes back through re-review after the fix. An edit carries the same fee as an approve. Reject declines with a documented reason. On every path your written rationale is the record; citing a guideline strengthens it, but no reference list is required. Every decision is timestamped and audit-trailed. See the full sequence
Every decision is recorded under your name. The pilot runs on synthetic cases and no review has been paid; the fee is set in writing before any live case.
Stay clinically engaged on your terms. Review from home, the cabin, or anywhere with WiFi.
Fill downtime between shifts with meaningful clinical review work.
Between assignments? Stay in clinical work without committing to a schedule. Pick up reviews when it suits you.
Build clinical decision-making experience while you launch your career.
Planned as one flat fee per review, the same whether you sign or decline. Not open yet: the pilot runs on synthetic cases, no review has been paid, and the fee is set in writing before any live case.
No minimums, no shifts. Review from anywhere — your couch, your office, between patients. Set your own availability once live cases open.
No entity-level malpractice, AI-liability, or cyber policy is in force today. Reviewers work under their own malpractice coverage, and the queue checks every case against that patient’s state and routes it only to a physician who is both licensed and covered there. Every other case in the queue is synthetic. If that changes, you will be told in writing.
Not measured yet. Until a review is completed under the current timing, this says so rather than showing an estimate.
Meaningful clinical review work on your own schedule.
Every approve, reject, and edit is recorded with your rationale, so what the draft got wrong is on the record.
Every case you signed or declined. No scores and no rankings.
Every review you sign generates a tamper-evident receipt: what you reviewed (by content hash — never the content itself), your verdict, your credential class, the timestamp. Those receipts accumulate into a portable, independently verifiable record of your reviews anchored to a public registry that is countersigned into Bitcoin daily, so it stays provable even without us. That record is yours by design: the reviews may be work for hire — the proof of how you judge is a professional asset you keep for the rest of your career.
Receipts carry no patient data — fingerprints only. The standard your reviews are measured against is public: The Five Tests of a Real Review.
No credentialing committee. No 90-day panel process. The fee is the same whether you sign, edit or decline, and it is set in writing before any live case. See the full sequence
ClinicalSwipe is built to protect reviewers, not just patients. Every safeguard exists so you can review with confidence.
No rubber-stamping. An approval with nothing in the case packet is refused on the server, and your time on each case is recorded — measured, not enforced.
AI output and original source data displayed together. You see exactly what the AI interpreted and the raw patient data it used.
Every decision timestamped, every edit version-controlled. Full forensic record if anyone ever asks.
In force today: each reviewer’s own malpractice policy, plus a queue that will not route a case to a physician who is not both licensed and covered where the patient is. Not yet in force: entity-level malpractice, AI-liability, or cyber. Until those bind, every case is synthetic or de-identified.
Clear scope of engagement. You review documents — you don't treat patients. Clean liability boundaries.
Enter your 10-digit NPI number. We verify against the CMS NPPES Registry in real time.