Configurable document fraud detection that adapts to your risk strategy

No two financial institutions evaluate fraud the same way. Every lender has different underwriting policies, document requirements, risk tolerances, and fraud patterns. Yet many document fraud detection solutions apply the same detection logic to every customer.

That one-size-fits-all approach creates unnecessary manual reviews while allowing institution-specific fraud patterns to go unnoticed. Teams end up adapting their processes to the software instead of configuring the software around the way they actually assess risk.

Inscribe's Decision Engine was built to solve that problem. It gives fraud and risk teams the flexibility to configure document fraud detection around their own workflows, allowing the platform to evolve alongside changing fraud patterns and internal policies.

Configure fraud detection around your institution

Inscribe includes a broad set of document fraud detectors out of the box, but every institution has unique requirements. The Decision Engine lets you tailor how those detectors behave while adding your own institution-specific rules. Rather than relying on static detection logic, you can continuously refine how documents are evaluated as your team learns more about the fraud patterns affecting your business.

Today, the Decision Engine allows you to:

• Tune detector severities
• Customize decision logic
• Manage software blocklists
• Create Custom Document Insights
• Create Custom Transaction Insights
• Configure document acceptance criteria

Instead of asking reviewers to remember dozens of internal policies, the Decision Engine applies them automatically across every application.

Automate your team's manual document reviews

Every fraud team develops manual checks over time. Some reviewers verify that bank statements were issued within the last 90 days. Others inspect document metadata for editing software, flag stitched statements, or check whether applicants submitted the correct document type.

These reviews are valuable because they reflect your team's experience. They're also repetitive. Custom Document Insights transform those manual checks into automated rules.

Using plain-language logic, your team can create institution-specific fraud checks that run automatically whenever a document is submitted. Whether you're validating document age, identifying screenshots, detecting metadata anomalies, or enforcing document requirements, those reviews become consistent across every application.

Instead of relying on reviewers to remember every policy, your expertise becomes part of the workflow.

Analyze bank statements using your own risk policies

Bank statements contain far more information than account balances. They reveal spending habits, income stability, cash flow, and financial behavior that often provide stronger indicators of risk than the document itself.

The challenge is that every institution evaluates those behaviors differently. With Custom Transaction Insights, your team defines exactly what should trigger additional review.

Build insights using transaction:

• Categories
• Merchants
• Payment processors
• Payment methods
• Dollar amounts
• Timeframes
• Counts or percentages

You can also evaluate:

• Average daily balances
• Balance volatility
• Minimum and maximum balances

For example, you might automatically flag applicants who:

• Spend more than 20% of their transactions on gambling or cryptocurrency exchanges.
• Maintain an average daily balance below your underwriting threshold.
• Show unusual balance volatility over a rolling 30-day period.
• Transfer unusually large amounts between personal accounts.
• Exhibit transaction patterns that don't align with your lending policies.

Before deploying a new rule, you can backtest it against historical applicants to understand exactly how often it would trigger and whether it improves detection performance.

Rather than relying on a generic fraud model, every bank statement is evaluated using your institution's own policies.

Put your fraud signals in context

Not every fraud signal should influence a decision the same way. Some indicators are highly predictive of fraud and should contribute directly to a customer's overall risk rating. Others provide useful context but aren't strong enough to affect automated decisioning.

The Decision Engine gives your team control over how every signal is used. You can:

• Adjust detector severities.
• Decide whether custom insights contribute to customer risk ratings or appear as contextual information.
• Apply detectors only to specific document types.
• Configure custom decision logic.
• Maintain blocklists for document editing software that should never appear in your application flow.

This flexibility allows teams to safely test new fraud rules, gradually introduce stricter policies, and continuously improve detection accuracy without disrupting existing workflows.

Detection that evolves with fraud

Fraud changes quickly. AI-generated documents continue to improve. New manipulation techniques appear every month. Your own fraud team constantly uncovers new behaviors that weren't part of your review process six months ago. Your fraud detection platform should be able to adapt just as quickly.

Instead of waiting for product releases or asking engineering teams to build one-off rules, the Decision Engine allows fraud teams to continuously refine how documents are reviewed, how signals are prioritized, and how customer risk is evaluated.

As your institution learns what predicts fraud within your own portfolio, that knowledge becomes part of every future review. The result is document fraud detection that becomes more precise over time — reducing false positives, identifying more true fraud, and reflecting how your organization actually evaluates risk.

If you'd like to see the Decision Engine in action, book a demo to learn how your institution can configure Inscribe around its own fraud strategy.