Paste a de-identified note; get the code line, the decision making behind it, and a paste-ready rationale.
Do not paste information from an EHR or clinical record unless all patient identifiers have been removed and your organization permits use of this service. Removing only the name, date of birth, and record number may not be sufficient: HIPAA Safe Harbor de-identification covers 18 identifier categories, including dates, detailed geography, phone numbers, email addresses, and device or record identifiers. You remain responsible for de-identification. The full checklist and privacy details. Pasted text travels encrypted to the AI provider solely to produce this analysis; submissions are never used to train models, and provider retention is limited, about 30 days under its published policy, before deletion. The raw note is never stored by this site; the analysis result is held briefly in temporary job storage while your browser retrieves it, deleted on pickup, and expired automatically within one hour either way. Educational support: code only what your documentation supports, per your organization's policy. Privacy details.
How to use this at work: keep this page open in a pinned tab beside your chart. Remove all patient identifiers from the note before pasting (see the warning above; your organization's policy governs), run it, then use the Copy button to paste the rationale back into your note before signing.
Every run is a live analysis of the note you paste. Values follow current CMS files.
What this tool does: a real AI model reads your pasted, de-identified office note; a deterministic rules engine then verifies the visit level under the 2021 MDM framework, applies modifier 25, 57, and 24 logic, checks G2211 and counseling add-ons, computes global period dates, and writes a paste ready rationale in your voice.
Who it is for: physicians and advanced practice clinicians who code their own outpatient visits: primary care, orthopaedics, hospital based clinics, and every specialty that reports 99202 to 99215. It reads real notes, including ambient AI drafts with heavy template text.
What you get: the supported CPT codes with wRVU values and a running total, the tier by tier MDM reasoning, honest advice on what one more documented sentence would support when the work was done, and a Copy button for the rationale.
How it was validated: a permanent battery of authored notes with known correct answers runs against the live tool after every release, each note in five noise wrapped variants; the deterministic engine is additionally cross validated against an independent implementation of the AMA rules across thousands of generated scenarios. Every confirmed miss becomes a pinned test. The full method.
Common questions: Is my note stored? No; nothing is stored by this site and submissions are never used to train models. Does it work for new patients, consults, and hospital visits? Yes; it selects the correct code family automatically. Does it handle time based coding? Yes; documented total time competes with MDM and the higher result wins. What about G2211? Eligibility is checked on every note and the attestation sentence is written for you. See also: 99213 vs 99214, G2211 documentation requirements, modifier 25 with a joint injection.
How the assistant works and how we tested it. Found a miscode? Report it; every confirmed miss becomes a pinned test.