Skip to main content
Back to challenges
Challenge

GAIP Climate Risk Insurance RAG Engine with LlamaIndex & Cognition Devin

Insurers must dynamically underwrite regional climate risks based on technical policies and historical flood data. Construct an advanced RAG parametric underwriting pipeline using LlamaIndex and Cognition Devin automation to generate validated climate policy decisions.

Workflow AutomationHosted by Vera
Sign in free to start
Challenge brief

What you are building

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

Build a climate insurance policy underwriting index with LlamaIndex and automated engineering with Cognition Devin.

Delivery guide

How work is evaluated

Evaluation

Evaluates policy decision accuracy and document retrieval correctness.

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

citation_present

Checks if clause citation exists

Binary check

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

Dimension 2

retrieval_accuracy

Correct policy clause retrieval rate • 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

  • Implement hierarchical RAG indexing using LlamaIndex

  • Automate code refinement using Cognition Devin environment execution

  • Extract structured underwriting recommendations from climate policy documents

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic policy guidelines and flood historical records for Southeast Asia.

Challenge setup

Evaluates policy decision accuracy and document retrieval correctness.

Challenge input

JSON query containing location and flood millimeter level

LlamaIndex

Document parsing and vector search

Cognition Devin

Automated pipeline engineering

Evaluated output

JSON containing payout_approved, payout_amount, citation

Evaluation checks
  • Checks if clause citation exists
  • Correct policy clause retrieval rate • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: ClimateRAG-Bench
  • Retrieval Accuracy 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
  • Cognition Devin · Optional
  • Exa · Optional
Observation
  • Cognition Devin · Optional
Orchestration
  • Cognition Devin · Optional

Evaluation contract

  • citation_present · Weight 1
  • retrieval_accuracy · Weight 1

Metrics

  • retrieval_accuracy · Target 0.95

Benchmarks

  • ClimateRAG-Bench
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON query containing location and flood millimeter level

    Output: JSON containing payout_approved, payout_amount, citation

    Execution harness: Python

Evaluation tools

  • LlamaIndex · RAG Framework
  • Cognition Devin · Agent System

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about GAIP Climate Risk Insurance RAG Engine with LlamaIndex & Cognition Devin