Upload the guest’s identity document and a vision agent reads it — NIC front and back, passport bio page, licence or residence card — and fills what it can read onto the booking.
SEE IT IN ACTION
Every field is nullable. The agent returns nothing rather than guessing, and a null never blanks what you typed.
OVERVIEW AND BENEFITS
The slowest thirty seconds of every check-in
Identity capture is required, error-prone and done while a guest stands waiting. Reading it from the document removes the typing without removing the operator from the loop.
Multi-page in one pass
NIC front and back, or both sides of a licence, are read together in a single extraction rather than one upload at a time.
Reads more than one script
Sinhala, Tamil, English and Latin text are all in scope, which is what a Sri Lankan front desk actually receives.
It never guesses
Any field the agent cannot read confidently comes back empty instead of plausible-and-wrong.
WHY IT MATTERS
A mistyped identity number is found months later, by an auditor
Manual entry of document numbers is where statutory guest records quietly go wrong — and the error surfaces long after anyone can check the original document.
4document types read
1pass for a multi-page document
0typed values overwritten by a blank
PRODUCT DETAILS
How it behaves on the desk
A vision agent behind an upload field, with the merge rules that make it safe to use during a live check-in.
Document extraction
Structured guest details read from an uploaded NIC, passport, driver’s licence or residence card.
Multi-page in one call
Several attachments collapse into a single extraction pass, so a two-sided NIC is one action.
Multi-script OCR
Reads across Sinhala, Tamil, English and Latin text on the same document.
Overwrite empty only
Extracted values fill blank fields; anything you already typed is left exactly as you typed it.
A vision model that guesses at a smudged digit produces an identity number that looks right and is not. This one emits null for anything it cannot confidently read.
Null means unread:the operator sees an empty field and knows to check the document.
Merge is additive:extraction fills blanks and never clears a value a person has entered.
The scan is kept:the original image stays on the booking, so the source of any field is still checkable.
COMPARISON
Vestibel vs. typing from the document
Capability
Vestibel
By hand
Scanner hardware
Reads NIC, passport, licence and residence card
Front and back in one pass
Sinhala, Tamil, English and Latin
Leaves unreadable fields empty
Scan archived on the booking
No extra hardware at the desk
CUSTOMER STORIES
“
Vestibel gave us one guest record across rooms, restaurant and spa — check-in time is down 40% and we stopped losing restaurant charges entirely.
Dinesh Weerakoon · General Manager, The Banyan House
Placeholder story — no customer-story content exists yet.
A 30-minute walkthrough with your rooms, rates and outlets — no slides.
FAQ
FAQ
Sri Lankan NIC front and back, passport bio pages, driver’s licences and residence cards, across Sinhala, Tamil, English and Latin text.
It is designed not to. Any field it cannot read confidently is returned empty rather than guessed, and the merge only fills blanks — so a null can never overwrite a value you typed.
The uploaded document is archived on the reservation’s identity-documents collection, so the source of every extracted field remains checkable.
Identity documents are one of the stores covered by the platform’s data-retention and subject-access handling, so they can be purged on a retention policy and reported on a subject-access request.
