Skip to main content
Loading navigation
Back to challenges

Build an Autonomous Digital Banking Onboarding Agent with Mastra AI and o3

Digital banks require real-time KYC verification and anti-fraud checks to safely onboard remote customers without adding friction. Build a Mastra AI workflow leveraging OpenAI o3 to perform cross-document entity verification and identity fraud detection, achieving under 2% false rejection rate.

CybersecurityHosted by Vera
Sign in free to start

What you are building

Implement an automated onboarding pipeline using Mastra AI workflows and o3 reasoning models to cross-verify document data against watchlist databases.

How work is evaluated

Evaluation

Evaluates the accuracy and speed of identity verification across synthetic KYC submissions containing fraud traps.

Datasets

Public datasets, plus the private datasets you have access to.

How submissions are scored

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

Synthetic Fraud Trap Rejection

Ensures mismatched identity records are properly rejected.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

False Rejection Rate

Percentage of valid users wrongly rejected. • target: 0.01 • range: 0-0.02

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

  • Design a stateful KYC onboarding workflow in Mastra AI with deterministic fallback steps.

  • Integrate OpenAI o3 reasoning capabilities for fine-grained identity document discrepancy analysis.

  • Implement real-time watchlist screening against sanctions datasets with entity resolution.

  • Maintain compliance traceability by outputting audit-ready decision logs.

Resources

Dataset notes

Synthetic database of Egyptian and international identity documents with embedded fraud patterns.

Challenge setup

Evaluates the accuracy and speed of identity verification across synthetic KYC submissions containing fraud traps.

Challenge input

JSON containing applicant profile, passport text, and proof of address string

Mastra AI

Orchestrates step-by-step KYC state flow in TypeScript.

OpenAI o3

Provides structured logical extraction over ambiguous text.

Evaluated output

JSON with verification status, risk_score (0.0 - 1.0), and reasoning trace

Evaluation checks
  • Ensures mismatched identity records are properly rejected.
  • Percentage of valid users wrongly rejected. • target: 0.01 • range: 0-0.02
Configured evaluation evidence
  • Benchmark: FintechKYCBench
  • False Rejection Rate target: 1%
  • 1 public reference case
Configured runtime
  • TypeScript execution harness
  • javascript sandbox (unavailable on Versalist)

Shows the first task, up to three tools, and up to three evidence items.

View technical recipe

Configured tools

Action Space
  • Mastra AI · Required
  • OpenAI · Optional
Policy Serving
  • o3 · Optional
Orchestration
  • Mastra AI · Required

Evaluation contract

  • Synthetic Fraud Trap Rejection · Weight 1
  • False Rejection Rate · Weight 1

Metrics

  • False Rejection Rate · Target 0.01

Benchmarks

  • FintechKYCBench
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing applicant profile, passport text, and proof of address string

    Output: JSON with verification status, risk_score (0.0 - 1.0), and reasoning trace

    Execution harness: TypeScript

Evaluation tools

  • Mastra AI · Agent Framework
  • OpenAI o3 · Reasoning Model

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build an Autonomous Digital Banking Onboarding Agent with Mastra AI and o3