Skip to content
Docklands by Inaza

AI fraud detection that checks the details.

Traditional systems validate data fields and run rules-based checks. They never look at the document itself. Docklands examines every claim and submission on intake — down to the pixel — and refers flagged files to your team with the evidence attached. Docklands surfaces the signal; your people decide.

Part of Solutions
Scope
Claims & submissions
Analysis
Down to the pixel
Decision
Your SIU
The problem

A rules engine cannot see a doctored image.

Field validation confirms a number is well-formed, not that the document it came from is real. The manipulation lives in the pixels, the metadata and the arithmetic — where nothing is looking.

0Blost to insurance fraud in the US each year
  • Validation is not verification

    Rules-based checks confirm a field is present and plausible. They never examine the document the field was read from.

  • Edits are invisible at volume

    A retouched invoice total or a cloned damage photo survives any review that moves at claim speed.

  • Synthetic documents arrive clean

    Generative tools produce a plausible, internally consistent repair estimate on demand — with no scanner artefacts to catch.

  • Manual review can’t cover the book

    Investigators screen a fraction of files. Soft fraud is paid without a second look because no one had time for a first one.

How Docklands solves it

Examine every file, on the same three moves.

Docklands reads every claim on intake, examines the document itself for manipulation, and refers risky files with the evidence — it surfaces risk, your investigators decide.

  1. 01 · Ingest

    Examine every file on intake

    Every claim and submission is captured and examined as it arrives — no file goes unchecked because no one had time.

  2. 02 · Analyse

    Check the document, not just the data

    Docklands inspects pixels, metadata and arithmetic for signs of manipulation, and verifies details against 100+ data sources — recording the evidence behind every flag.

  3. 03 · Refer

    Refer to your SIU

    Flagged files route to your investigators with the indicators and source data attached. Docklands surfaces the risk; your people make the call.

Out of the box

Add document forensics in 15 minutes.

Map your fraud indicators and SIU thresholds onto a review agent, connect your verification sources, and examine every file from day one.

  1. 01Define your signals
  2. 02Set referral thresholds
  3. 03Connect data sources
  4. 04Go live
Capabilities

Manipulation, caught in the details.

  • 01

    Pixel-level tamper detection

    Retouched totals, cloned regions and splice boundaries surfaced from the image itself — not from the fields read off it.

  • 02

    Metadata forensics

    Capture timestamps, GPS coordinates, device fingerprints and edit history read from the file and checked against the claim.

  • 03

    AI-generated document detection

    Synthetic invoices, estimates and receipts identified by the artefacts generative models leave behind.

  • 04

    Mathematical verification

    Line items, subtotals and totals recomputed across every document — arithmetic that does not reconcile is flagged.

  • 05

    Physical tampering signals

    Reprints, overlays, inconsistent lighting and re-photographed screens detected before extraction runs.

  • 06

    Evidence trail, your SIU decides

    Every flag records the indicators behind it against your own thresholds. Docklands never declines a claim — your investigators do.

How it works

Examining the whole book, not just a sample.

  • 0%

    Of claims screened on intake

  • 0/7

    Continuous screening

  • 0:1

    Rules mapped to your SIU playbook

  • 0+

    Decision reasons recorded

  • Docklands flags; your SIU decides
  • Evidence attached to every referral
  • Region pinning US · EU
Common questions

Document fraud detection, answered.

How does AI fraud detection work for insurance claims?+

Docklands examines every claim document down to the pixel rather than validating the data fields on it. It combines tamper detection, metadata forensics, AI-generated document detection and mathematical verification, then refers what it finds to your SIU with the evidence attached.

Why do rules-based checks miss document fraud?+

Because they validate what a document says, not whether the document is real. A rules engine cannot see that an invoice total was edited in Photoshop, that a receipt’s metadata was written after the loss date, or that the whole document was generated by a model.

Can it detect AI-generated documents?+

Yes — that is a first-class check, not an add-on. AI-generated invoices, receipts and estimates are now a routine attack, and they are indistinguishable from real ones to a rules engine.

Does Docklands decide whether a claim is fraudulent?+

No. Docklands flags and evidences; your SIU decides. Referral thresholds map 1:1 to your existing SIU playbook, and every referral arrives with the evidence trail behind it.

How much of the book gets screened?+

100% of claims, on intake, 24/7 — the whole book rather than a sample. Every screened file records 200+ decision reasons.

What is Docklands by Inaza?+

Docklands is Inaza’s document-forensics product for insurance. It runs on the same platform as the rest of Inaza, so screening happens on intake alongside claims and submissions automation.

Expedited Start

Examine your claim documents this week.

Structured PoC in 4 weeks — no onboarding cost. Run Docklands against your own claims and keep every decision with your team.