VaultMint
On-deviceAES-256INR
Indian personal finance, without bank linking

Track every rupee, automatically.

Hand it a bank SMS or a statement PDF. It reads the rows, resolves the merchant, assigns the category and catches the duplicates — all on your own computer.

No bank login. No account linking. No cloud account.

Reads bank SMSParses statement PDFs and email receiptsTurns them into budgets and trends
Ask for a walkthrough

No signup, no hosted demo, no cloud sync.

From clutter to clarity

One message, start to finish

Bank SMS → transactionLive
HDFCBK · 14 Jul, 8:42 pmRs.1247.50debitedfrom a/c XX4821on 14-07-26to VPA swiggy.payu@hdfcbankUPI Ref 451203876512. Not you? Call 18002586161 -HDFC Bank
Amount−₹1,247.50
MerchantSwiggyswiggy.payu@hdfcbank
Date14 Jul 2026
AccountHDFC ····4821
CategoryFood & Dining
Confidence0.94
1

Messages become transactions

SMS, email receipts and PDF statements are read on-device and parsed into clean, categorised entries.

Parse
2

Transactions manage your budgets

Each entry routed to the right budget, with the remaining balance updating as it lands.

Insights

What the numbers are actually telling you

Spending by category

How each category moved across the last three months

MayJunJul

Net cash flow

Money in against money out, last six months

InOut

Credit utilisation

How much of each card limit is in use

Food−8% from June
Shopping+71% month over month
Axis Magnus56% utilised

Sample data — not a real account

Features

The core product, not a feature dump

The homepage shows the pieces that make VaultMint feel distinct.

Duplicate protection

Re-import a statement and nothing doubles — while ₹500 withdrawn twice in one day still counts twice.

ATM · 14 Jul · 18:04 ₹500 kept
ATM · 14 Jul · 21:37 ₹500 kept
Statement re-upload 184 rows dropped

Bank SMS

Alerts from HDFC, ICICI, SBI, Axis, Kotak, Yes, IndusInd, IDFC First, Federal and PNB — and not limited to them.

Statement documents

Bank and credit card PDFs, including password-protected ones, with live extraction progress.

Confidence, not silence

Every extraction is scored. Anything uncertain waits in a review queue with the original text beside it.

Swiggy 0.94
Amazon 0.88
TPT/4471 0.41

Auto-categorisation

Swiggy to Food, Uber to Transport — with Indian categories like gold, government services and household help.

Email inbox

Bank emails become transactions too. Statement attachments are routed through the document pipeline.

Every account type

Not one generic account. Each type carries the detail that actually matters to it.

SavingsCurrentOverdraftSalary NRENROCredit cardLoanMF folio

Recurring detection

Finds the ₹499 subscription you forgot — same merchant, same amount, same day.

Insights dashboard

Category trends, net cash flow, balances, credit utilisation, bills due and fee-waiver progress.

Privacy by design

It stays with you, because there is nowhere else for it to go

Most money apps ask you to hand over bank credentials. VaultMint cannot — it never connects to your bank.

SMS / PDF / Email
Local model
Encrypted store
Dashboard
No hosted account, analytics, or bank-link server

Read on your machine

Statements are read locally, not sent to a provider.

No servers to hold it

No hosted VaultMint service holds transaction history.

Encrypted where it rests

Sensitive fields use AES-256 before storage.

The key is yours alone

It never enters the database. No reset, no override.

Questions

Security and privacy, in plain English

Can anyone else read my transactions?
No. There is no VaultMint server for your data to reach — your copy runs entirely on your own machine and contacts nothing. The sensitive fields are also encrypted with a key that exists only on that machine.
What exactly is encrypted?
Merchants, descriptions, notes, account numbers, review-queue text and extracted document content are encrypted before they are written to storage, using AES-256-GCM authenticated encryption. The product keeps amounts, dates and category codes readable so it can still sort, total and chart them without decrypting every row.
How can it search encrypted data at all?
VaultMint does not search the ciphertext directly. For fields like merchant name it stores a blind index: an HMAC-SHA256 hash of the normalised plaintext, keyed with that user's own encryption key. That allows exact-match and prefix search without exposing the original text, and the same merchant hashes differently for different users.
Where is the encryption key stored?
Not in the database. The master key is meant to live in the machine's secure key store or an explicit secret source, and it wraps a separate per-user data key. That separation is why losing the key is fatal: there is no hidden recovery copy sitting in the app's tables.