Challenge

Construct an Anti-Scam Fraud Triage Crew with CrewAI & Resemble AI

Public sector anti-scam units face escalating phone and voice impersonation attacks. Construct a CrewAI multi-agent workflow integrated with Resemble AI voice verification to analyze reported scam interactions, categorize fraud tactics, and output structured triage reports with a false positive rate under 3%.

Workflow AutomationHosted by Vera
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build a CrewAI multi-agent anti-scam triage workflow using Resemble AI synthetic voice analysis to identify deepfake voice scams.

Delivery guide

How work is evaluated

Evaluation

Evaluates the accuracy of scam classification and false positive rate in detecting synthetic voice impersonation attacks.

Datasets

Shared data for this challenge

Review public datasets and any private uploads tied to your build.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

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

triage_priority_match

Checks if high-severity deepfake scams are assigned HIGH priority.

Binary check

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

Dimension 2

false_positive_rate

Rate of legitimate calls misclassified as scams. • target: 0.01 • range: 0-0.03

Binary check

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

Learning goals

What you should walk away with

  • Build role-based multi-agent teams using CrewAI for incident triage

  • Integrate audio forensic signals from Resemble AI into agent tools

  • Classify scam operational tactics according to national scam taxonomies

  • Produce automated actionable alert summaries for law enforcement agencies

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic multi-modal dataset containing 100 call transcripts paired with Resemble AI voice verification scores.

How this agent runs

Evaluates the accuracy of scam classification and false positive rate in detecting synthetic voice impersonation attacks.

Challenge input

JSON containing transcript and Resemble AI audio forensic scores

CrewAI

Facilitates multi-role collaborative agent workflows.

Resemble AI

Provides synthetic audio detection scores for voice scam triage.

Evaluated output

JSON triage report with scam taxonomy and priority score

Checks for
  • Checks if high-severity deepfake scams are assigned HIGH priority.
  • Rate of legitimate calls misclassified as scams. • target: 0.01 • range: 0-0.03
Proof of success
  • Benchmark: ScamDetectionBench
  • False Positive Rate target: 1%
  • 1 public reference case
Runtime evidence
  • Python execution harness
  • Python sandbox (unavailable on Versalist)
View technical recipe

Configured tools

Action Space
  • CrewAI · Required
  • crewAI · Optional
  • Resemble AI · Optional

Evaluation contract

  • triage_priority_match · Weight 1
  • false_positive_rate · Weight 1

Recipe state

This is a preview. The configuration can change before the evaluation recipe is locked.

Frequently Asked Questions about Construct an Anti-Scam Fraud Triage Crew with CrewAI & Resemble AI