Skip to main content
Back to challenges
Challenge

LlamaIndex AI Underwriting Governance & Auditability Agent

Regulatory frameworks require insurers to maintain transparent, auditable records of all automated model underwriting decisions. Develop a LlamaIndex RAG governance pipeline that tracks decision lineage and generates regulatory audit reports for AI underwriting models.

Machine LearningHosted by Vera
Sign in free to start
Challenge brief

What you are building

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

Use LlamaIndex dynamic indices and lineage trackers to audit model inference histories and regulatory compliance.

Delivery guide

How work is evaluated

Evaluation

Evaluates audit trail completeness and accuracy in identifying non-compliant or biased model decisions.

Datasets

Challenge datasets

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

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

audit_flag_check

Ensures non-compliant decision is flagged correctly

Binary check

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

Dimension 2

audit_precision

Precision of regulatory compliance flagging • target: 0.95 • range: 0-1

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 a governance logging index in LlamaIndex tracking model prompts and completions

  • Generate automated compliance reports verifying non-discrimination standard adherence

  • Implement lineage querying to explain individual model underwriting decisions to regulators

Resources and assets

Reference links and supporting material

Dataset notes

Inference audit log files from automated underwriting systems containing inputs, features, and quotes.

Challenge setup

Evaluates audit trail completeness and accuracy in identifying non-compliant or biased model decisions.

Challenge input

JSON containing batch underwriting inference logs

LlamaIndex

Provides callback handling and vector indexing over model interaction logs

Evaluated output

JSON with audit_passed (bool), compliance_score (0-100), and flagged_decision_ids

Evaluation checks
  • Ensures non-compliant decision is flagged correctly
  • Precision of regulatory compliance flagging • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: AI-Governance-Bench
  • Audit Precision target: 0.95
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Llama Index · Required
  • xAI · Optional
  • Exa · Optional

Evaluation contract

  • audit_flag_check · Weight 1
  • audit_precision · Weight 1

Metrics

  • audit_precision · Target 0.95

Benchmarks

  • AI-Governance-Bench
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing batch underwriting inference logs

    Output: JSON with audit_passed (bool), compliance_score (0-100), and flagged_decision_ids

    Execution harness: Python

Evaluation tools

  • LlamaIndex · Agent & Indexing Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about LlamaIndex AI Underwriting Governance & Auditability Agent