Skip to main content
Loading navigation
Back to challenges

Office-to-Residential Structural Conversion Assessment Pipeline with AI SDK & Haystack

Commercial conversion projects encounter severe budget overruns when structural floorplate limitations are identified late in planning. Build a web application with Vercel AI SDK and Haystack to extract core depths and window distances from architectural floor plans. Achieve 95% metric extraction precision on structural blueprints.

Business OperationsHosted by Vera
Sign in free to start

What you are building

Build an AI SDK dynamic TypeScript tool with Haystack document pipeline integration to grade office conversion suitability from architectural PDF data.

How work is evaluated

Evaluation

Evaluates accuracy of extracted floorplate dimensions and conversion feasibility grading.

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

feasibility_boolean_check

Checks if feasibility flag accurately matches expected grade based on core depth threshold

Binary check

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

Dimension 2

extraction_precision

Precision of structural numeric parameters extracted • 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

  • Use Vercel AI SDK (`ai` and `@ai-sdk/openai`) to generate structured output for architectural metric extraction

  • Leverage Haystack pipelines for document indexing, chunking, and semantic searching across building engineering specs

  • Implement client-side interactive UI with React and streaming output for feasibility scoring

  • Validate extracted core-to-window depth against municipal residential daylighting requirements

Resources

Dataset notes

Dataset of technical building specification text excerpts describing office floorplates, core shafts, and column spacings.

Challenge setup

Evaluates accuracy of extracted floorplate dimensions and conversion feasibility grading.

Challenge input

JSON containing spec string, depth dimensions, and window counts

AI SDK

TypeScript AI SDK for building streaming AI interfaces.

Haystack

Document pipeline framework for searching architectural engineering specs.

Evaluated output

JSON containing core_to_window_distance_ft, conversion_feasibility, yield_grade

Evaluation checks
  • Checks if feasibility flag accurately matches expected grade based on core depth threshold
  • Precision of structural numeric parameters extracted • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: Architectural Document AI Benchmark
  • Extraction Precision target: 0.95
  • 1 public reference case
Configured runtime
  • JavaScript 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
  • Haystack · Required
  • Haystack · Optional
  • Dagster · Optional

Evaluation contract

  • feasibility_boolean_check · Weight 1
  • extraction_precision · Weight 1

Metrics

  • extraction_precision · Target 0.95

Benchmarks

  • Architectural Document AI Benchmark
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing spec string, depth dimensions, and window counts

    Output: JSON containing core_to_window_distance_ft, conversion_feasibility, yield_grade

    Execution harness: JavaScript

Evaluation tools

  • AI SDK · agent_framework
  • Haystack · rag_framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Office-to-Residential Structural Conversion Assessment Pipeline with AI SDK & Haystack