Product · Document fraud checks

Catch tampered documents before you approve.

Lenders, insurers and finance companies approve on documents: IDs, bank statements, salary slips, invoices and claim photos. We check every one for signs of editing and inconsistency, and send anything suspicious to your team with the reason attached.

What we've delivered

BFSI Document fraud checks built for a financial services client: ID validation, edited-image detection and PDF metadata tampering, with analysts reviewing every flag.

What it checks

Six ways a document gives itself away.

Each document in an application or claim goes through every check that applies to it.

01

ID documents

Passports, national ID cards and licences: format, expiry, machine-readable zone check digits, and whether the details match the rest of the file.

02

Edited images

Photos and scans of documents: signs of local editing such as pasted text, cloned areas, and mismatched compression or resampling.

03

Tampered PDFs

Bank statements, salary slips and invoices: edits made after the file was issued, the software that last saved it, and fonts or text layers that don't belong.

04

Numbers that don't add up

Running balances that don't reconcile, totals that don't match their line items, and salary credits that don't match the payslip.

05

Cross-document mismatches

Name, date of birth, address, employer and account numbers compared across every document in the file.

06

Reused documents

The same document, photo or template turning up across different applicants or claims.

How it works

Checks run in the background. People make the call.

  1. 01

    Documents arrive

    Uploaded to your loan or claims system, emailed in, or sent by API.

  2. 02

    Read and sorted

    Each file is identified as an ID, statement, payslip, invoice or photo.

  3. 03

    Checked

    Every check that applies runs on every page, and documents are compared with each other.

  4. 04

    Flagged with reasons

    Each document is marked clear or review, with the specific reason and the area highlighted.

  5. 05

    Your team decides

    Nothing is rejected automatically. Analysts review the flags and make the call.

What your team sees

A reason, not just a score.

A flag your analysts can't explain is a flag they'll ignore. Every flag says what was found, where, and why it matters, so review takes minutes instead of a fresh look at the whole file.

Example · Application 2291
salary_slip_march.pdfReview

Edited 9 days after it was created. The net salary figure uses a different font from the rest of the page.

bank_statement_q1.pdfReview

Closing balance on page 3 doesn't follow from the transactions listed.

passport_scan.jpgClear

Machine-readable zone valid. Name and date of birth match the application.

Honest limits

What it won't do.

Make the decision for you

It narrows what your analysts look at. Approvals, rejections and fraud referrals stay with your team.

Catch every fraud

A genuine document carrying false information, or a forgery printed and re-scanned, can pass. We test on your past cases and tell you what it caught and what it missed.

Replace your KYC or bureau checks

It adds a document-level layer alongside the identity verification, credit bureau and sanctions checks you already run.

Questions

Before you book.

Which documents can it check?

IDs, bank statements, salary slips, utility bills, trade licences, invoices and claim photos are the common ones. We start with the two or three document types where you lose the most money or time.

How do we know it works on our documents?

Before anything goes live we run it on a sample of your past files, including known fraud cases where you have them, and report what it flagged, what it missed and how many false alarms it raised.

Where are documents processed?

We agree where documents are processed and stored before anything is built, and work within your security, data-residency and compliance requirements.

Does it slow down approvals?

Checks run as documents arrive, so files that come back clear move on without waiting. Your team spends its time on the ones that are flagged.

How long until it's live?

The AI Opportunity Audit takes about two weeks. The first checks are usually live within a four-week Implementation Sprint.

How many documents does your team check by eye each week?

In a 45-minute discovery call we look at where fraud and manual checks cost you most, and whether these checks would catch it.

Book a discovery call

45 minutes · $30, credited to your Audit