Vision Proofread
Vision Proofread checks your translation on the rendered page, catching what only shows up once it is laid out: a heading that overflows, a CTA that lost its emphasis, text that no longer fits its frame. It looks at the page the way a reader would.
When should I use Vision Proofread instead of Smart Proofread?
Vision Proofread is for HTML documents — full-body emails and pages — where layout matters as much as wording. It uses the model’s native vision to review the rendered result and flag visual or layout issues alongside the usual language checks.
Like Smart Proofread, it applies its fixes inline as they land, with progress shown in a toast. For plain text documents, Smart Proofread is the tool you want.
Can I run Vision Proofread on a Word document or a PDF?
No — Vision Proofread runs on documents imported as HTML, where there is a rendered page for the model to look at. For Word, PDF, Markdown, and plain-text documents, Smart Proofread is the pass to use; it runs the same language checks, without the visual layer.
Do I still need to run Smart Proofread after a Vision Proofread?
No. Vision Proofread runs the usual language checks alongside the visual ones — it is the same fix flow as Smart Proofread with visual evidence added, so one pass covers both.
How do I undo a change Vision Proofread made?
Vision Proofread applies its fixes straight to the active translation as they land, but nothing is lost: each fix is saved as a new version of that block. Open the block’s version history and switch back to the earlier version.
Still stuck? Ask Literess in the app, or write to [email protected].