← Back to Home

Paul & The Future of CAD

Two Industries, One Question: Is AI Coming for Design Work?

About Paul

My brother Paul designs data centers for AI using Revit and Civil 3D. The irony isn't lost on us: he's designing the physical infrastructure that houses the AI systems that might eventually replace parts of his workflow.

Paul's first lesson: "AutoCAD is dead. We compete against Revit and Civil 3D." He's right - AutoCAD is legacy 2D drafting. The real tools are BIM (Building Information Modeling) systems that understand buildings as intelligent 3D objects, not just lines.

The Parallel Between Our Industries

In my industry, the transformation has already happened. I no longer write code in the traditional sense. Instead, I:

My Current Role (Software)

  • Conceive and articulate ideas
  • Define requirements and architecture
  • Supervise AI-generated code
  • Review, approve, and refine outputs
  • Focus on strategy and creative problem-solving

Paul's Current Role (BIM/CAD)

  • Model structures in Revit (BIM)
  • Design site/infrastructure in Civil 3D
  • Apply building codes and regulations
  • Coordinate mechanical/electrical/plumbing systems
  • Generate construction documents and schedules
"The question isn't whether AI can do design work. The question is: when will the workflow shift from manual BIM/CAD modeling to AI-supervised generation?"

The Future Scenario

Imagine This Workflow (5-10 Years?)

1. Natural Language Input "I want to build a 3,000 sq ft house in Arizona. 4 bedrooms, 3 baths, open concept kitchen and living room. High ceilings, lots of natural light. Modern desert aesthetic. Must comply with Pima County building codes."
2. AI Generates Multiple Concepts AI produces 5-10 different architectural concepts with 3D renders, floor plans, and basic structural analysis. Takes into account climate, local codes, site orientation, and aesthetic preferences.
3. Iterative Refinement "I like option 3, but can you move the master bedroom to the east side and add a covered patio?" AI regenerates instantly with modifications.
4. Approval & Detailed Generation Once approved, AI generates complete construction documentation: detailed floor plans, elevations, sections, structural calculations, MEP (mechanical/electrical/plumbing) layouts, material specifications.
5. Export to Standard Formats Output to DXF, DWG, IFC, or whatever format contractors and building departments need. No CAD program touched by human hands.

In this scenario, CAD programs like AutoCAD and Revit become intermediate formats that AI writes directly to—not tools that humans manipulate. The designer's role shifts from drafting to creative direction and quality control.

Why This Seems Inevitable

The Technology Already Exists - LLMs can understand complex requirements. Image generation models can create photorealistic renders. Parametric design tools can generate optimal structures. The pieces are all there.
Building Codes Are Codifiable - Unlike creative taste, building codes and engineering constraints are well-defined rules. Perfect for AI to learn and apply.
Economic Pressure - Just like in software, the cost and time savings will be impossible to ignore. A design that takes weeks could take hours.
The Software Pattern - My industry transformed in about 2-3 years (2022-2025). CAD could follow a similar trajectory once the tools mature.

What Changes, What Stays

What Likely Changes:

Manual BIM modeling disappears. No more spending hours moving walls in Revit, adjusting grading in Civil 3D, or coordinating MEP systems. AI handles all the tedious, repetitive work.

Software skills become less valuable. Knowing Revit or Civil 3D inside and out matters less. The new skill is knowing how to effectively communicate design intent to AI and evaluate its outputs.

Iteration speed increases dramatically. Try 10 different layouts in a day instead of laboring over one for a week.

What Stays Critical:

Design vision and aesthetic judgment. AI can generate options, but someone still needs to know what looks and feels right. Paul's understanding of what makes a good data center layout—airflow, cable management, maintainability—that's not going away.

Domain expertise and practical knowledge. Understanding how buildings actually get built. Knowing what contractors can and can't do. Anticipating problems before they happen.

Client communication and project management. Understanding what the client actually wants (often different from what they say). Managing expectations. Coordinating with contractors and engineers.

Open Questions

  • Liability and licensing: Who signs off on AI-generated plans? Do architects and engineers still need to stamp drawings? How do professional licensing boards adapt?
  • Building department acceptance: Will building departments accept AI-generated plans? What verification processes need to be in place?
  • Complexity ceiling: Can AI handle truly novel or unusual projects? Or will it work great for standard structures but struggle with unique challenges?
  • Timeline: Is this 5 years away? 10 years? Or has it already started and we just haven't noticed?
  • The role of CAD companies: Does Autodesk pivot to AI-first workflows? Or do they get disrupted by startups building design-to-DXF tools from scratch?

My Take

Having lived through this transformation in software, I think it's when, not if. The shift will probably be faster than people expect, just like it was in my industry.

The survivors will be the ones who embrace it early. Not the people with the best AutoCAD skills, but the people with the best design judgment and the ability to effectively direct AI. Just like in software, the winners aren't the fastest typists—they're the people who understand what to build and can evaluate whether it's correct.

Paul designs the infrastructure for AI. Soon, AI might return the favor and handle the BIM modeling work for him. That's not a threat—that's evolution. It means he gets to spend more time on the interesting problems: creative design, client needs, and complex engineering challenges. Less time fighting with Revit and Civil 3D.

The irony is beautiful. And the future is probably closer than we think.

But Wait... Is This an Opportunity?

The more interesting question: if Autodesk is so big and dominant, but too slow to pivot to AI-first workflows... is there a disruptive opportunity here?

What if a small team—someone who understands AI and software architecture, and someone who deeply understands BIM/CAD and design workflows—built the tool that bypasses traditional BIM tools entirely?

Read the Business Plan →