Drop the documents. The reading is done for you.
Aadhaar, PAN, degree certificates, licences, offer letters. Drop them in together and each one is read on the spot — no queue, no overnight job. The employee confirms their own documents; HR verifies. What used to be a week of typing becomes a review.
Bulk In, Structured Out
Drop a folder of mixed documents. Each is identified, read and filed against the right person with its number and dates extracted.
Two-step Verification
The employee confirms their own documents are correct; HR verifies. Neither step is skippable, which is what makes the record defensible later.
Expiry That Surfaces
Documents with dates — licences, contracts, visas — carry their expiry, so the lapse is visible before it becomes a problem.
Encrypted At Rest
Identity numbers are stored encrypted, and every export of personal data is written to the audit trail.
Bulk in, structured out
The usual onboarding pattern is a form with fourteen upload fields and a person who types the contents of each document into the field beside it. The documents are already the data; retyping them is the waste.
Drop the whole set in together. Each file is read on the spot — no queue, no overnight batch — and the fields it contains are proposed against the employee's record. What arrives as a folder of photographs leaves as structured, checkable data.
Two steps, because they are different questions
The employee confirms that the reading is right: this is my PAN, that is my degree certificate. HR then verifies that the document is acceptable: it is legible, it is current, it is the document we asked for.
Keeping those apart matters. A verified document and a merely submitted one are not the same thing, and treating them as one is how organisations discover at audit that half the vault is unchecked.
Expiry that surfaces itself
Licences, registrations, contracts and visas expire. A vault that stores them without watching the dates is a filing cabinet. Expiry dates are captured at reading time and surface before they matter, rather than on the day somebody is stopped at a gate.
From a folder of photographs to a verified file
Drop them all
Several documents at once — photographs, scans, PDFs, whatever the employee actually has.
Each is read
Inline, at the moment of upload. Fields are drafted and the document type identified.
The employee confirms
They check their own details. This is the reading being confirmed, not the document being approved.
HR verifies
A separate act, by a different person, recorded separately — with expiry captured for whatever expires.
Document Vault FAQs
What document types does it read?
The documents Indian employment actually runs on — Aadhaar, PAN, driving licences, certificates, offer and appointment letters, and the miscellany a joining file accumulates. Anything the AI cannot read confidently is flagged for a person rather than guessed at, because a confident wrong identity number is far more expensive than an honest blank. Bulk drops work: each file is read, matched and drafted individually.
Is the AI reading ever wrong?
Sometimes — which is exactly why the flow has two human checkpoints. The employee confirms what was read from their own documents, and HR verifies before it becomes the record; the AI removes the typing, not the checking. Uncertain values arrive flagged, so attention goes where it is needed instead of everywhere. The failure mode is a correction on screen, never a silently wrong record.
Where are the files stored, and how securely?
In your own tenant on our infrastructure in India, behind the same isolation boundary as the rest of your data. The sensitive fields extracted from them — bank account, IFSC, national identity and travel-document numbers — are additionally encrypted at rest at the application level, on top of transport encryption. Every view and export of a document is written to the audit trail.
Who can see an employee's documents?
The roles you allow, and the employee themselves — and the trail records every look. That last part changes behaviour more than any permission matrix: "who opened this employee's Aadhaar last month" is a question your administrator can answer with a name and a timestamp, which is precisely what makes holding identity documents defensible at all.
How does bulk collection actually work at onboarding?
Drop the pile — a season's intake, a migration from the old filing cabinet — and the AI reads each file, matches it to a person and drafts the record. Employees confirm their own via self-service before day one; HR verifies rather than transcribes. Document collection usually fails at the typing step, and this removes exactly that step while keeping both human checks.
Can we get the documents back out?
Yes — the export-everything commitment includes the files themselves, not just the data read from them, at any time and without asking. Documents your employees gave you are part of your record; a vault you cannot empty on your own decision would be a trap with a friendly name, and it would contradict everything the security page promises.
See it on your own data
Fourteen days, no card. Load a handful of employees and judge it against how you work today.