Turn a Photo of Handwriting Into Editable Text

Not a dedicated one — and we won't pretend otherwise. @vustScribeBot, VUST's transcription bot, is audio/video-in only: send it a voice note or a recording and it returns a transcript and a summary. It does not take a photo and has no OCR pipeline. A dedicated handwriting-to-text tool (photo of a page → clean editable text) is not built. The waitlist button on this page is a real demand counter — enough votes and it gets built.

Because it does a different job. Scribe converts spoken audio into text (a meeting recording, a voice memo) — that's speech-to-text. Handwriting-to-text is a different technical problem: reading pixels on a page (varied handwriting styles, ink quality, page skew) and converting that into structured text, closer to document OCR than to speech transcription. VUST doesn't have that pipeline today for either bot.

Clean, editable digital text that mirrors the structure of the page: paragraphs kept as paragraphs, bullet or numbered lists kept as lists, and — where useful — a plain confidence note on any word the model wasn't sure about, rather than silently guessing. The target is 'ready to paste,' not a raw dump of recognized characters.

That's the real test of any handwriting-OCR tool, and it's also the honest limit: legibility varies enormously by handwriting style, pen, lighting and photo angle, and no tool — VUST's proposed one included — will read genuinely illegible handwriting perfectly. A well-built tool should flag low-confidence words rather than invent text, but accuracy on very messy cursive is a real constraint worth setting expectations for up front.

Access shape, mainly. Most handwriting-OCR tools live on a website: create an account, upload the image, download the result. A Telegram-native version would mean sending the photo to a bot you already have open and getting text back in the same chat — no new account, no separate upload page. That convenience is the wedge; it wouldn't claim to out-recognize a dedicated OCR engine on hard cases.

Only informally: @vustbot is general multi-model AI chat, and depending on which model answers in a given session it may be able to read text out of an image you send as part of a conversation. That's not a dedicated, consistent handwriting-to-text product with formatting rules — it's general chat capability that varies. If a reliable, purpose-built version matters to you, the waitlist vote is what moves it up the roadmap.