Every year, Chicago YIMBY publishes a countdown of the tallest projects under construction. In 2021, I had an idea: what if I built all of them in 3D and revealed one per day?

I spent the month adding each tower to a downloaded base model of Chicago in Unity. Every article included screenshots showing that day's project (highlighted in cyan) among all the previously revealed towers (shown in green), set against a grayscale Chicago basemap.

The response was immediate. Readers loved seeing spatial relationships - how the new towers compared to existing landmarks, where clusters were forming, what the skyline would look like when everything was complete.

This was validation that spatial context matters when understanding urban development.

The December countdown worked, but it was limited - one view, one color scheme, one story. To make this truly useful, I needed multiple perspectives on the same data.

I experimented with different classification schemes:

The goal was to answer different questions with the same spatial data:

• "Where is construction most active?" → Status view
• "What building types dominate this neighborhood?" → Function view

This required rethinking the entire color system and building a data architecture that could support multiple classification layers simultaneously.

I had a hypothesis: people who grew up playing SimCity would intuitively understand this interface. But my early models looked clinical - like data diagrams rather than explorable environments.

Jakob's Law states that users prefer interfaces that work like things they already know. I needed to bridge the gap between analytical tool and familiar experience.

The solution: make it look like a game.

Natural building colors instead of status indicators. Realistic textures suggesting glass, concrete, brick. Environmental context - parks, water, infrastructure. Camera angles that balanced verticality with spatial relationships.

The goal was familiarity.

By late 2022, I was using the model not just for visualization, but for advocacy. Data presented visually makes arguments that text cannot.

Three projects exemplified this shift:

1. COMPLETE CITYSCAPE: Applied natural colors to ALL buildings over 200 feet, not just in-development projects. Full context shows what's changing vs what already exists.
2. ALTERNATIVE SCENARIOS: Modeled canceled mega-projects to show "what if?" Chicago has a history of ambitious proposals that never got built. Seeing them in context sparked conversations about opportunity cost.
3. COMPARATIVE ANALYSIS: Extracted Chicago's tallest under-construction towers and placed them side-by-side with New York's to make a visual argument about scale and ambition.

The model had become a rhetorical device.

What if Chicago's most ambitious canceled projects had actually been built? I used the model to visualize never-built mega-developments, showing readers what the skyline would look like today if these proposals had proceeded.

Two views told the complete story: natural colors to show visual impact, status layer (in red) to emphasize their hypothetical nature.

At the end of 2022, both Chicago YIMBY and New York YIMBY ran their annual countdown of tallest projects under construction. I saw an opportunity for visual storytelling.

I extracted the top 15 from each city and arranged them side-by-side by ranking - starting with Chicago's #15 next to New York's #15, adding the next tallest pair each day until reaching #1 on the final day of the countdown. The Willis Tower, Eiffel Tower, and Great Pyramid of Giza appear alongside the lineup as scale references.

The image made an argument without words: New York (orange) is building at a dramatically larger scale than Chicago (green), America's supposed "second city" of skyscrapers.

This visualization helped readers understand the scale gap between the two markets and sparked discussions about Chicago's regulatory environment and development pipeline.

In Spring 2023, I published a comprehensive breakdown of West Loop's development boom. Rather than covering projects individually (my usual approach), I analyzed the entire neighborhood holistically - breaking down the pipeline by height ranges.

The article used the model extensively: multiple angles, different data layers, and an "x-ray effect" showing which buildings fell into each height category.

Reader response was overwhelming. People didn't just want to read about developments - they wanted to explore the data themselves. They wanted filters, sorting, export capabilities.

This was product-market fit validation. People would pay for this.

By mid-2023, my co-founder and I had partnered with YIMBY Network - the parent organization behind Chicago, Miami, and New York YIMBY. This partnership gave us:

1. Access to their comprehensive project databases
2. Built-in distribution through their established readership
3. Brand validation and credibility

But raw data wasn't enough. Project information lives in hundreds of articles, each with slightly different formats and structures. We needed a way to extract structured data from unstructured text.

The solution: an AI-powered extraction pipeline.

The biggest technical challenge was data acquisition. Development projects are announced through news articles, not structured databases. How do you turn natural language into queryable data at scale?

We built an extraction system that:

1. Ingests article text from YIMBY publications
2. Uses LLMs to identify key fields (height, address, developer, architect, status, units, square footage, etc.)
3. Validates extracted data against known patterns and ranges
4. Reconciles conflicting information across multiple sources
5. Outputs clean JSON for the platform database

This is the same challenge faced by any company processing unstructured documents - legal contracts, insurance policies, medical records. The architecture is broadly applicable beyond real estate.

While Unity models were powerful, they had limitations: download requirements, WebGL performance issues, poor mobile experience. We needed a web-native dashboard.

The platform we built provided:

• Interactive map with project pins and clustering
• Filterable, sortable data tables with advanced queries
• Analytics and aggregations (by year, status, neighborhood, developer)
• Real-time updates as new projects are added to the database
• Export functionality for researchers and professionals
• User accounts with saved searches and notifications

If the platform worked for Chicago, would it work elsewhere? Between June and November 2023, we launched three cities to find out:

• CHICAGO (June 2023) - Home market, 20 data layers plus the 1960-2033 timelapse
• MIAMI (September 2023) - Test in a different urban context (waterfront focus)
• NEW YORK (November 2023) - Largest dataset with 5,000+ projects and 200+ of the largest rendered in 3D

Each launch validated the platform in a new market. New customers continued to validate the product by subscribing for $40 a month or signing up for yearly subscriptions.

As the platform grew, users started asking questions geared towards market intelligence:

• "Which developers are most active in this neighborhood?"
• "What has this architecture firm designed recently?"
• "Who are the dominant general contractors?"

This required expanding our data model from tracking buildings to tracking stakeholders and their relationships. Every developer, architect, contractor, and owner mentioned in our database became a queryable entity.

The result: instant industry analysis.

Unity provided powerful 3D rendering but had critical limitations:

• Required WebGL (performance-heavy, desktop-only)
• Difficult to integrate with standard web components
• Poor mobile experience

We needed a hybrid approach: Unity's visual impact with web-native accessibility.

The solution: Mapbox GL JS with GLTF building overlays.

By 2025, our user base had evolved. We weren't just serving urban enthusiasts - real estate professionals, city planners, researchers, and policy analysts were subscribing.

These users needed more than visualizations. They needed:

• Official regulatory data (zoning, permits, building codes)
• Advanced filtering with saved queries
• Flexible viewing modes for different contexts
• Professional-grade performance at scale

This phase focused on serving power users without alienating casual ones - enterprise features with consumer-grade UX.

We integrated directly with New York City's official property database (NYC DOB), pulling comprehensive regulatory data:

• Building Identification Numbers (BIN)
• Borough-Block-Lot (BBL) tax codes
• Zoning district classifications (R7-2, C6-4, M1-5, etc.)
• Community district assignments
• Fire company jurisdictions
• Lot area and building dimensions
• Owner records and assessed property values
• Certificate of Occupancy status

This transformed YIMBY+ from a visualization tool into a comprehensive research platform.

As the platform matured, we refined our data layers from 19 to 20, removing redundant or problematic attributes while adding more valuable dimensions:

Removed:

• "Completion Decade" - Less granular than Completion Year
• "Neighborhood" - Geographic boundaries too inconsistent across sources
• "Primary Function" - Created confusion when Project Type already existed

Consolidated:

• "Floors" + "Stories" merged into "Number of Floors" - Eliminated metric redundancy

Added:

• "Primary Material" - Construction material data (glass, concrete, brick) for sustainability analysis
• "Total Units" - Comprehensive occupancy metrics
• Three stakeholder layers: "Developer", "Architect", and "General Contractor" - Project team accountability and portfolio tracking

The final 20 layers span physical attributes, project details, and stakeholder intelligence. But with this much dimensionality, we faced a UX challenge: how do you provide granular control without overwhelming users?

The solution: a vertical slider that lets you explore data layer values sequentially - like scrubbing through a timeline. Simple interaction, powerful results.

Different contexts demand different aesthetics:

• LIGHT + COLOR: Default mode for general browsing and exploration
• LIGHT + MONOCHROME: Focus on building form, reduce visual noise for analysis
• DARK + COLOR: Optimized for low-light environments and evening use
• DARK + MONOCHROME: Presentation mode with maximum contrast

Rather than choosing one aesthetic, we built all four - then let users switch instantly.