Part 2 From Potential to Practice: Navigating AI Across the Design Process - Pre-Design

Before any form is sketched, architects must navigate the complex terrain of pre-design—where client goals, budgets, site constraints, and regulations converge. AI’s value here isn’t in rendering, but in organizing and clarifying information. Tools leveraging natural language processing and simulation can help synthesize briefs, analyze code, and test feasibility—but only when matched to the specific demands of early-phase work. In this post, we explore three such platforms: one for rapid feasibility, one for AI-assisted programming, and one for automated code review.

Forma: Rapid Feasibility Study

Designed for early-stage massing and performance analysis, Autodesk’s Forma allows architects to test basic site strategies and quickly visualize potential yields. It’s particularly effective for developers and firms working under tight pre-design timelines—offering solar analysis, wind simulations, and floor area estimates with minimal input. But Forma’s power comes with tradeoffs: sketches lack precision, and transitioning its outputs into Revit or Rhino often requires rebuilding geometry. It’s not a replacement for modeling, but a tool to surface possibilities and support fundraising or client pitches—especially when site data is scarce and time is limited.

TestFit: Programming

Built as a generative planning engine, TestFit excels at quickly generating and optimizing building layouts based on program requirements, parking constraints, and zoning rules. With intuitive inputs (unit mix, target yield, site shape), it creates test fits in seconds—ideal for multifamily developers, housing authorities, or firms conducting rapid feasibility studies. Its strengths lie in real-time iteration and visual clarity, but it performs best when the design brief is already well defined. Integrating with Revit and AutoCAD is possible via exports, though data loss and cleanup are often required. TestFit augments programming—it doesn’t replace the need for a designer’s insight into adjacency or spatial narrative.

UpCodes: Code Review

Regulatory constraints often shape projects before design begins, and UpCodes AI brings the power of NLP to code compliance. Architects can query local zoning and building codes in plain language—What’s the minimum corridor width for a preschool?—and receive targeted responses. For small firms without in-house code experts, this is a major time-saver. It works within web browsers, integrates lightly with BIM workflows, and delivers solid accuracy for standard use cases. Still, it’s not a substitute for judgment: edge cases, historic overlays, or client-specific compliance strategies require human expertise. It augments research, but doesn’t replace legal or regulatory review.

AI can’t replace the essential conversations or critical thinking that define the pre-design phase. But it can help architects arrive at those conversations better prepared—armed with analysis, data, and early options. The key is matching the right tool to the right task. For developers, that might mean using Forma or TestFit to test massing strategies or layouts. For small firms, tools like UpCodes AI can accelerate research and reduce risk. As we move forward in this series, we’ll continue to assess each tool’s core functionality, who it’s built for, how it integrates (or doesn’t), and what it leaves unsolved. Because when it comes to AI in architecture, clarity is power—and pre-design is where clarity begins.