01
Answers in the group
Customers post a reference or a receipt photo; the agent replies in the thread with the verified status, amount and webhook state.
Everypayment,verified inseconds.
Transaction-verification agents for fintech support groups
Built forPayment gateways+Wallets+Merchants+Reconciliation teams+Support desks
01 / The live inbox
One queue for
every group.
The agent answers the references it can verify. Everything else — not found, pending, mismatched, risky — lands in one realtime inbox with an SLA, and your reply goes back to Telegram under the customer's message.
- 01
Read
Telebirr and M-Pesa references, platform IDs, receipt links, photos and PDFs — confusable characters repaired.
- 02
Verify
Checked against your payment platform in real time: status, amount, merchant and webhook delivery.
- 03
Answer
A reply in the thread in seconds, or a case for your team with everything already looked up.
02 / Fraud & scammers
Catch the scam,
explain why.
Twenty-one explainable detectors score every claim before anyone is paid. Risky ones are held for a reviewer — in the dashboard or with buttons in Telegram — and every decision teaches the model.
- 01Inflated amounts and edited receipts
- 02References replayed, probed or reused across merchants
- 03Velocity and fan-out from one account
- 04Risk spread from confirmed scammers through the entity graph
03 / The connected workspace
Built for the whole
operation.
02
Reads receipts like a person
Screenshots, camera photos, dark-mode captures and PDFs. Receipt links are trusted over smudged print.
03
Scammer intelligence
An entity graph links phones, accounts and receipts — and, if you opt in, privacy-preserving signals shared across businesses.
04
Exceptions with SLAs
Not found, pending, amount and merchant mismatches, webhook failures and risk reviews each get their own queue and timer.
05
Know your operation
Auto-resolution rate, response latency, volume per group and agent performance — live.
06
API and webhooks
Verify from your own systems with scoped keys and receive signed events as they happen.
Your team should handle the exceptions,
not every reference.
04 / Live in an afternoon
From sign-up to
answering.
- 1
Connect
Add your Telegram bot or support account, and your payment platform's API.
- 2
Map your groups
Tell us which merchant each group belongs to. Every transaction is checked against it.
- 3
Go live
Run in shadow mode to compare first, then let the agent answer — your team handles only the exceptions.
05 / Questions
Before you start.
Which payment platforms work?
LakiPay out of the box, and any platform with an HTTP lookup API through the configurable connector — status, amount, merchant and webhook fields are mapped once in Settings.
Does it need a bot, or can it use our support account?
Either. A Telegram bot is simplest; a logged-in support account works in groups where you can't add a bot. Both send through the same rate-limited outbox.
What happens when it isn't sure?
It doesn't guess. Unreadable receipts, lookups that fail and risky claims become cases for your team, with the evidence attached, and the customer is told a person will check.
Can we try it without customers seeing anything?
Yes. Shadow mode composes every reply and shows it in the dashboard without sending it, so you can compare with your current process before switching on.
Where is our data, and who can see it?
Each workspace is isolated with row-level security in the database; bot tokens and API secrets are envelope-encrypted, and phone numbers are stored as keyed pseudonyms.
Put an agent in
every group.
Free to set up. Shadow mode first, live when you're ready.