Skip to main content
Back to challenges
Challenge

Office-to-Residential Conversion Feasibility Engine with Mastra AI & Hume AI

As commercial real estate faces structural office vacancies while urban markets demand residential housing, office-to-residential conversions have emerged as a key urban renewal strategy. However, developers frequently hit structural realities: floor plate depths exceeding 100 feet hinder natural lighting, plumbing stack alignments fail residential codes, and complex zoning regulations stall approvals. In this challenge, you will construct an intelligent conversion feasibility and stakeholder engagement system using Mastra AI in TypeScript. You will build a Mastra AI workflow that ingests floor plans, column grid spacings, core-to-glass distances, and municipal zoning PDFs to evaluate residential layout conversion viability. To streamline executive and investor presentation workflows, you will integrate Hume AI to provide an empathetic voice-assisted discovery interface. The interface allows asset managers to query conversion bottlenecks (e.g., lightwell necessity, HVAC retrofitting costs) and receive real-time structural risk summaries with tailored vocal tonality based on project risk severity.

Business OperationsHosted 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 an office-to-residential conversion evaluation workflow using Mastra AI agent workflows in TypeScript and Hume AI voice assistant integration.

Delivery guide

How work is evaluated

Evaluation

Evaluates the Mastra AI conversion workflow on structural parameter assessment, zoning compliance flag detection, and overall response accuracy.

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

lightwell_threshold_check

Verifies that floor_plate_depth_ft > 90 correctly triggers lightwell_required = true.

Binary check

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

Dimension 2

valid_cost_range_check

Ensures est_retrofit_cost_psf falls within industry standard range ($150 - $450/sqft).

Binary check

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

Dimension 3

workflow_execution_accuracy

Accuracy of extracted building metrics and generated conversion feasibility scores. • 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 Mastra AI workflow with dynamic step dependencies to evaluate structural conversion metrics

  • Extract window-to-core depth, ceiling height, and MEP plumbing stack data from unstructured building documents

  • Integrate Hume AI Expressive Voice & Empathic API for natural voice interactive status reporting

  • Compute automated spatial conversion feasibility scores and estimated retrofit costs per square foot

Resources and assets

Reference links and supporting material

Dataset notes

Dataset of 50 commercial office building spec sheets (floor plate sizes, window distances, structural grid details, HVAC types) across NYC, Chicago, and San Francisco along with municipal zoning requirements.

Challenge setup

Evaluates the Mastra AI conversion workflow on structural parameter assessment, zoning compliance flag detection, and overall response accuracy.

Challenge input

JSON with building_id, gross_sqft, floor_plate_depth_ft, ceiling_height_ft, existing_hvac_type, and zoning_district

Mastra AI

TypeScript-native agent workflow engine for executing multi-step business logic.

Hume AI

Provides empathic voice synthesis and expressive audio interfaces for executive reporting.

Evaluated output

JSON with conversion_score, lightwell_required, max_residential_units, est_retrofit_cost_psf, and audio_briefing_prompt

Evaluation checks
  • Verifies that floor_plate_depth_ft > 90 correctly triggers lightwell_required = true.
  • Ensures est_retrofit_cost_psf falls within industry standard range ($150 - $450/sqft).
  • Accuracy of extracted building metrics and generated conversion feasibility scores. • target: 0.95 • range:...
Configured evaluation evidence
  • Benchmark: CRE-Conversion-Feasibility-Bench
  • Workflow Execution Accuracy target: 0.95
  • 1 public reference case
Configured runtime
  • JavaScript execution harness
  • javascript 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
  • Hume AI · Required
  • Mastra AI · Optional
  • Merge · Optional
Orchestration
  • Mastra AI · Optional

Evaluation contract

  • lightwell_threshold_check · Weight 1
  • valid_cost_range_check · Weight 1
  • workflow_execution_accuracy · Weight 1

Metrics

  • workflow_execution_accuracy · Target 0.95

Benchmarks

  • CRE-Conversion-Feasibility-Bench
  • materials

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON with building_id, gross_sqft, floor_plate_depth_ft, ceiling_height_ft, existing_hvac_type, and zoning_district

    Output: JSON with conversion_score, lightwell_required, max_residential_units, est_retrofit_cost_psf, and audio_briefing_prompt

    Execution harness: JavaScript

Evaluation tools

  • Mastra AI · Agent Systems / Workflow Framework
  • Hume AI · Assistants & Interfaces

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Office-to-Residential Conversion Feasibility Engine with Mastra AI & Hume AI