Who you're paying
Is this someone you've paid before, or a brand-new payee? Has the community flagged this UPI ID?
Catch the scam
before you tap send
WallUPI is a payment app with one extra superpower. It runs a real-time risk check on every transaction and steps in before the money's gone.
Spots scams in real time · Works on-device · No real money moves
Built to do one thing brilliantly: stop you from paying a scammer.
3-tier
AI working together on every payment
On-device
Your messages are read on your phone
Real time
Checks run the moment you hit pay
Community
Flagged UPI IDs warn everyone
The problem
Scams don't win because people are careless.
They win because they're fast.
1
A panic SMS arrives: "your account is blocked."
2
A "bank officer" calls and stays on the line.
3
You're rushed into sending money to "fix" it.
4
The moment you confirm, it's gone.
WallUPI is the voice in your corner, pointing out the warning signs before you tap send.
Every payment, screened
Five signals, weighed in an instant
The moment you hit pay, WallUPI quietly reads the situation, not just the amount.
Is the amount normal
A sudden, out-of-character transfer to a new contact stands out, even when everything else looks fine.
Are you on a call
Scammers keep you talking so you can't think. An active call during payment raises the alarm.
A suspicious SMS
If a scam-like message recently mentioned this recipient, WallUPI treats it as a serious warning.
How you got here
Did you scan a QR or follow a link? Those entry points carry more risk than typing an ID yourself.
One clear verdict
All the signals come together into a single, easy-to-read risk read-out, so the decision is yours and fully informed.
Under the hood
Three layers of AI, working together
They run side by side on every message and payment: fast, private, and always on.
1
On-device · always on
Instant rule engine
A lightning-fast first pass that runs entirely on your phone, with no internet needed. It catches the obvious red flags the instant they appear.
2
On-device ML · private
TinyBERT on your phone
A compact language model reads messages locally and sorts them into scam categories: phishing, fake KYC, OTP theft and more. Fast, offline, and your data never leaves the device.
3
Cloud second opinion
A larger model, when it counts
When something looks borderline, a sanitized, anonymous snippet is sent to a powerful cloud model for a closer look, returning a plain-English explanation of what it found and why.
What you'll see
Four levels of intervention
The riskier it looks, the harder WallUPI works to make you pause.
Safe
Looks normal. Your payment goes straight through, no friction.
Soft warning
Something's a little off. A quick review card spells out the specific reasons.
Strong warning
Several red flags at once. A full warning screen, and you confirm you understand before proceeding.
Critical
This looks like a live scam. A mandatory cooldown pauses the payment and breaks the pressure before you can continue.
See it in action
A payment app that looks out for you
Familiar on the surface, quietly vigilant underneath.
Home
Everything in view, risk included
Your balance, recent contacts and quick actions, exactly like the apps you already use. The difference sits right at the top: a live banner that flags a suspicious SMS the moment it lands, so you're warned before you ever start a payment.
Send money
The recipient is checked as you type
Enter a UPI ID or number and WallUPI instantly pulls up any scam-like message tied to it. Here a "your account is BLOCKED, update KYC" text is linked to the payee, flagged at 92% risk before you tap Continue.
Risk warning
When it's critical, the payment pauses
A 100% risk reading stops the transfer and shows you why: the recipient, the amount, an active call, a suspicious SMS, broken down factor by factor. The forced pause is the point: it breaks the pressure a scammer is counting on.
Message analysis
See exactly why a message is a scam
Open any flagged message and WallUPI lays out its reasoning: the fraud category, a clear risk score, and the specific signals it found, from false urgency and a fake KYC request to a suspicious link or a money-transfer ask.
Inbox scan
Your whole inbox, read on-device
The native detector scans every message on your phone and tags it: 92% fake KYC, 96% lottery scam, 5% genuine bank alert. Real messages stay quiet; the dangerous ones get marked, so nothing slips past you.
Privacy first
Your data stays yours
Protection shouldn't cost you your privacy. WallUPI does the heavy lifting right on your device.
- ✓ Your messages are read and analysed right on your phone, never uploaded.
- ✓ If a borderline message needs a second opinion, only an anonymised snippet leaves the device. Names, numbers and account details are stripped first.
- ✓ Your transaction history stays on your device.
- ✓ Report a scam UPI ID once, and everyone in the community is warned.
Shape WallUPI
Help us build a safer way to pay
Answer the following questions to help us build something that keeps you safe!
Take the surveyGood to know
Frequently asked
What exactly is WallUPI?
A UPI payment app with a scam check built in. Every time you pay, it reads the situation: the recipient, the amount, your recent messages and whether you're on a call, then steps in before the money leaves your account. It's a demonstration app, so no real money moves, but the detection underneath is real.
How does it catch a scam I can't see?
It weighs five signals at once: who you're paying, whether the amount is unusual for you, whether you're on a call right now, whether a scam-like SMS recently mentioned this number, and how you reached the payment screen. Together they produce a single risk score from 0 to 100.
Will it block my payment, or just warn me?
It depends on the score. Low risk goes straight through. Medium risk shows a quick review card. High risk asks you to confirm you understand the warning. Critical risk pauses the payment with a forced cooldown, enough time to break the pressure and think again.
It reads my SMS. Is that private?
Yes. Messages are analysed on your device, not in the cloud. Only when something looks genuinely borderline is an anonymised snippet sent for a second opinion, with names, numbers and account details removed first.
Does it need an internet connection?
No. The on-device rule engine and the TinyBERT model handle the everyday checks completely offline. The cloud model is only consulted for the rare borderline case.
How is this different from my current UPI app?
Most apps move the money the instant you confirm. WallUPI adds a safety layer that looks for fraud first, so a scam is caught before you tap send, not reported after the money's already gone.