Head-to-Head Architecture Comparison
When clinicians evaluate documentation systems, the visible output is a note, but the cost sits elsewhere: in the minutes between a finished draft and a signed chart. This review compares Merry AI and Suki AI on three measurable axes for multi-specialty groups: annual software cost, the path text takes into the EHR, and how defensible the resulting documentation is under payer review.
Suki's architecture is vendor-integrated. Its public developer documentation describes Ambient APIs and Web SDKs in which the partner application owns the UI and decides how results move into the EHR. Its integrations with Epic, Oracle Health, athenahealth, and MEDITECH are deep and well built, and they are a genuine strength in large single-EHR enterprises. Merry AI takes a different path. Its Chrome extension maps EHR fields by ARIA labels and section context, then writes the reviewed note directly into the native textarea.
| Feature | Merry AI | Suki AI |
|---|---|---|
| Published pricing | $54/mo (Pro Annual) | ~$299 to $399/mo reported list |
| Annual cost per clinician | $648 | ~$3,588 to $4,788 |
| DOM injection speed | Direct field write, typically under 1 second | Send step or vendor integration |
| Group note-splitting | One note per consented participant | Not publicly documented |
| CPT G2211 capture | Signal-based, editable justification | E/M coding API for partners |
| Attestation logging | Audio to edit to signature chain | Platform audit logs |
| Data retention | Audio discarded after finalization by default | Per enterprise BAA terms |
The Cost Math, Line by Line
Suki does not publish a single universal price. Publicly reported per-provider list pricing has ranged from roughly $299 to $399 per month. These plans are usually sold on annual agreements, and enterprise terms are negotiated separately. Two other vendors give useful context. Freed lists at $99 per month. Nuance DAX Copilot has commonly been contracted near $8,000 per clinician per year.
At the upper reported tier, Suki costs about $4,788 per clinician annually. Merry AI Pro Annual at $54 per month is $648. The difference is $4,140 per clinician, an 86% reduction. For a 12-physician multi-specialty group, that is $49,680 per year before any change in coding capture. At the lower $299 tier, the gap is still $2,940 per clinician, or 82%.
Where the Hidden Friction Accumulates
Copy-paste is not free time. Suppose a clinician spends 40 seconds per encounter moving text between a review panel and the chart. Across 22 encounters a day, that adds nearly 15 minutes of daily overhead, or about 60 hours a year. Merry's direct field write removes that transfer step. The clinician reviews in place, edits in place, and signs in the native EHR workflow. Nothing is auto-signed.
Multi-Specialty Clinical Depth
Multi-specialty groups need specialty logic, not one generic template stretched across departments. Merry's note engine carries versioned criteria libraries for each department:
- Behavioral health: DSM-5-TR criteria clusters and PHQ-9 or GAD-7 scores.
- Cardiology: NYHA class and LVEF phenotype.
- Orthopedics: Joint-specific ROM degrees and MRC strength grades.
- Pediatrics: Bright Futures periodicity elements.
Each library writes into the matching section of the chart rather than one undifferentiated block.
Behavioral Health and Group Sessions
Group psychotherapy is where diarization becomes a billing question. A CPT 90853 session with six participants requires six individual notes, each documenting that patient's participation and progress. Merry separates speakers, attributes statements only to consented participants, and drafts one note per patient for clinician review. Suki's public documentation does not describe a comparable group note-splitting workflow, so test it directly during evaluation.
Procedures, Modifier 25, and 59
Same-day procedures need structural separation. When an E/M visit and a joint injection occur together, Merry drafts the evaluation and decision-making in one section. The procedure narrative goes in another, with indication, consent, technique, and findings. Distinct anatomical sites stay in distinct segments to support Modifier 59 review. Modifier language is suggested conservatively and flagged for clinician confirmation. It is never inserted silently.
G2211 Capture and Audit Defensibility
CMS describes G2211 as an add-on for visits where the practitioner serves as the continuing focal point for care or manages a single serious or complex condition over time. It is reportable with office/outpatient codes 99202 to 99215 and with home or residence visits. CMS's May 2026 MLN guidance addresses how G2211 interacts with Modifier 25, so review your billing rules against current manuals.
The recovery math is direct. G2211 pays roughly $16 per claim. Four qualifying established visits per clinic day across 245 days yields about $15,600 per physician annually. Merry detects longitudinal complexity signals, such as multi-drug reconciliation, specialist coordination, and social determinants. When it finds them, it drafts an editable justification sentence. When the signals are absent, it drafts nothing.
Every inserted segment is logged. The Clinical Logic Bridge records the audio window, model version, DOM insertion timestamp, and each clinician edit before signature. It then appends an attestation block confirming human review. That chain also supports the disclosure and human-in-the-loop expectations emerging in state AI statutes. Peer-reviewed literature on ambient documentation accuracy is indexed at https://www.ncbi.nlm.nih.gov/pmc/ for teams building an evidence file.
Data retention follows your policy. By default, Merry discards raw audio after note finalization. It retains text and logs according to the schedule your practice configures.
How to Run Your Own Evaluation
The fairest test is local. Pick one week, three specialties, and ten encounters in each. Measure the seconds from encounter close to signed note, and count the manual transfer steps. Then have your coder score G2211 and modifier documentation without knowing which vendor produced it. If Suki wins in your Epic environment, that is useful information. Schedule a 15-Minute Workflow Audit and we will review the chart together.


