Quick answer
What to do
In the current vetverve beta, staff identify the patient and visit on the paper, then upload a clear photo or scan. If a vision provider is connected, vetverve can turn the handwriting into editable text. A person must compare that text with the original and correct it before adding it to the SOAP note. The responsible clinician then reviews and signs the note as usual; vetverve never adds or signs the extracted text automatically.
Use a paper page that fits how the veterinarian works
Paper does not have to mean an unstructured blank page. Print a clinic-approved sheet with patient identifiers, visit date, author, and generous S, O, A, and P areas. Keep prompts light enough that the veterinarian can write in the order that works during the visit.
Assign each page to one patient and one encounter before writing begins. If additional pages are used, number them and repeat the patient identifier. Avoid loose scraps that require someone later to infer which chart they belong to.
Capture a faithful, readable original
Photograph or scan the entire page on a flat, contrasting surface with even light. Check all corners, faint pencil, margin notes, check marks, and page numbers before leaving the capture step. A technically successful upload can still be unusable if a shadow covers a dose or the bottom of the plan is cropped.
Keep the original image associated with the draft so the reviewer can move between source and extraction. Follow the clinic's privacy, retention, and device-handling procedures when a phone or scanner is used; convenience does not remove those obligations.
- Patient and visit identifiers are visible and match the open chart
- The full page and every margin are in frame
- Text is upright, in focus, and free of glare or shadow
- Multi-page notes have an obvious order
- The original remains available during review
Use handwriting extraction only to create a draft
Handwriting extraction in the current beta requires a connected vision service. Without one, staff can still store the page and type from it manually. The screen also shows whether extraction is processing, complete, failed, or unavailable.
Extracted text opens as a draft for review, never as part of a signed chart. Keep the S, O, A, and P sections separate, but check abbreviations, decimals, units, negatives, medication names, and text that may have landed in the wrong section.
Compare the draft with the page before applying anything
Place the original and extracted draft close enough to compare line by line. Review patient identity first, then dates, S/O/A/P placement, numbers, decimal points, units, medication details, diagnostic terms, negations, and follow-up timing. Mark uncertain text for deliberate resolution rather than smoothing it into a plausible sentence.
A person must compare the extracted text with the source and correct every error before adding it to the SOAP draft. Adding the reviewed text is a separate action; uploading a page or completing extraction does not change or sign the note.
Record who reviewed the text and when it was added
After adding the reviewed text, return to the clinic’s normal SOAP process. The clinician responsible for the record should check the complete note, remove any remaining placeholders, and sign only when it meets clinic policy and local requirements.
Retain enough provenance to show that a handwritten source existed, who reviewed the extraction, and when text entered the note. The clinic should set retention, access, correction, and destruction procedures based on its own professional and legal requirements. Vision extraction alone does not establish record compliance.
Make the failure path visible and recoverable
Plan for illegible writing, incomplete pages, unsupported files, unavailable vision configuration, low-confidence output, and a patient mismatch. Staff should be able to stop extraction, preserve the source, type from the original, or ask the author to clarify without losing the rest of the note.
Review a sample of completed conversions with the clinicians who use them. If the same abbreviation, layout, or capture problem repeats, adjust the paper template or scan instructions. Do not hide recurring uncertainty behind silent automatic correction.
Sources and scope
How this guide was prepared
Vision output can look plausible and still be wrong. The clinic’s authorized reviewer remains responsible for the final record. The clinic must also follow local requirements for source documents, confidentiality, corrections, and retention. The source pages below were checked July 12, 2026.
- 1Artificial Intelligence Risk Management Framework: Generative AI ProfileNational Institute of Standards and Technology / checked July 12, 2026
- 2Model Regulations: Medical RecordkeepingAmerican Association of Veterinary State Boards / checked July 12, 2026