Challenge

Automate BPO Insurance Claims Intake with AI SDK and Cognition Devin Agent

Offshore BPO processes require high-precision automated intake workflows to prevent claims processing errors. Build a streaming claims intake application using Vercel AI SDK and simulated Cognition Devin coding agent automation. Achieve under 2 seconds end-to-end processing latency per claim with 95% automated extraction precision.

Machine LearningHosted by Vera
Challenge brief

What you are building

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

Build a TypeScript/Node.js claims processing microservice using Vercel AI SDK integrated with Cognition Devin task workflows.

Delivery guide

How work is evaluated

Evaluation

Evaluates TypeScript AI SDK streaming pipeline speed and field extraction accuracy.

Datasets

Shared data for this challenge

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

Loading datasets...
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

latency_check

Total processing time per document is less than 2000 milliseconds.

Binary check

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

Dimension 2

field_extraction_f1

F1 score across extracted structured fields • 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 Vercel AI SDK `generateObject` and `streamObject` methods.

  • Integrate Cognition Devin capabilities for code generation and workflow setup.

  • Build type-safe claims schema extraction using Zod in TypeScript.

  • Measure stream latency and schema validation accuracy.

Resources and assets

Reference links and supporting material

Dataset notes

Dataset of 300 unstructured customer call logs and incoming emails for insurance claims BPO intake.

How this agent runs

Evaluates TypeScript AI SDK streaming pipeline speed and field extraction accuracy.

Challenge input

Unstructured text string representing intake call report

AI SDK

Vercel standard framework for type-safe TypeScript AI apps.

Cognition Devin

Automates setup, testing, and continuous deployment workflows.

Evaluated output

Validated Zod JSON object with caller_name, loss_date, loss_type, estimate

Checks for
  • Total processing time per document is less than 2000 milliseconds.
  • F1 score across extracted structured fields • target: 0.95 • range: 0-1
Proof of success
  • Benchmark: StructuredExtractionBench
  • Field Extraction F1 target: 0.95
  • 1 public reference case
Runtime evidence
  • JavaScript execution harness
  • javascript sandbox (unavailable on Versalist)
View technical recipe

Configured tools

Action Space
  • Cognition Devin · Required
  • Aim · Optional
  • Seldon Core · Optional
Observation
  • Cognition Devin · Required
Orchestration
  • Cognition Devin · Required

Evaluation contract

  • latency_check · Weight 1
  • field_extraction_f1 · Weight 1

Recipe state

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

Frequently Asked Questions about Automate BPO Insurance Claims Intake with AI SDK and Cognition Devin Agent