Building My Personal Website with AI: A Multi-Model Workflow

Last week I set out to rebuild my personal website. What would have traditionally taken weeks of back-and-forth with a developer—or months of learning to code myself—took one focused session. The result is what you're looking at right now.

This post isn't about the website. It's about the workflow: how I used multiple AI tools together, learned to prompt effectively on the fly, and discovered that the biggest barrier for non-technical builders isn't the technology—it's just getting started.

The Approach: Treating AI as a Creative Collaborator

I didn't start with a design in mind. I started with a reference.

Instead of trying to describe what I wanted, I found a website I admired and asked Claude to review it. I gave it a link to a site showcasing modern design aesthetics and asked it to analyze what made the design effective—the color palettes, typography choices, layout patterns, and interactive elements.

This changed everything. Rather than me fumbling to articulate "I want it to feel professional but not corporate," Claude extracted the design principles from real examples. Dark backgrounds with earthy accent colors. Subtle scroll animations. Glass-effect navigation. Gradient overlays on images.

The Power of Reference-Based Prompting

Here's what I learned: AI tools are dramatically more effective when you give them something concrete to work from. Instead of abstract descriptions, give them:

Examples: "Look at how this site handles the hero section"

Constraints: "Use only these three colors"

Comparisons: "Make it feel more like X and less like Y"

The AI went from generic suggestions to highly specific implementations that matched my taste—because I had shown it my taste rather than trying to describe it.

Working Across Models: Playing to Strengths

Here's where it gets interesting. While Claude was building the website, I realized I needed a logo. Claude is exceptional at code and reasoning, but I wanted to explore visual concepts. So I switched to ChatGPT with its image generation capabilities.

I described what I was looking for—something that represented my military background (a shield) combined with forward momentum (an ascending arrow). After a few iterations, I had concepts I liked. Then I used another tool to convert the image into clean SVG code, which I brought back to Claude to integrate into the site.

This multi-model workflow felt natural once I stopped expecting one tool to do everything.

AI Model Strengths: A Practical Comparison

Based on this project and others, here's how I think about different models:

The key insight: don't be loyal to one tool. Use the right model for each task and move outputs between them freely.

The Real-Time Iteration Loop

What surprised me most was the speed of iteration. In a traditional development process, each change request means waiting—for a developer to find time, understand the request, implement it, and deploy it.

With Claude Code, the loop collapsed to seconds.

"The name should be all gradient, not just the first name": → Done in 20 seconds

"Can you make the Affirm tile purple to match their brand?": → Done

"I added a King Air image to the folder, use that for the Army section": → Updated immediately

This isn't just faster—it's a fundamentally different creative process. When iteration is instant, you explore more options. You try things you wouldn't have bothered requesting from a developer. The gap between imagining something and seeing it shrinks to almost nothing.

The Favicon Debugging Saga

Not everything was smooth. At one point, I couldn't get my custom logo to show up as the browser favicon. We tried multiple approaches—different file formats, different locations, clearing caches, incognito mode.

The problem? There was an old icon file in a directory that was overriding the new one. Classic development debugging—but with AI, even the troubleshooting was collaborative. Claude methodically worked through possibilities, explained what each attempt was testing, and eventually identified the conflict.

This is the part that would have frustrated me to no end if I were trying to learn coding on my own. With AI, debugging becomes a conversation rather than a solo struggle through Stack Overflow.

Getting Started: GitHub and the Intimidation Factor

I want to address something that almost stopped me before I started: GitHub.

If you've never used version control, GitHub can feel intimidating. Branches, commits, pull requests—it sounds like a foreign language. I nearly decided to "just make changes directly" and skip the whole thing.

I'm glad I didn't. Here's why it matters:

Safety net: Every change is tracked. If something breaks, you can always go back

Clean workflow: Working in branches means you can experiment without fear

Professional practice: If you ever work with developers, this is how they work

Claude walked me through the workflow: create a branch, make changes, commit with a message, push, create a pull request. After doing it twice, it felt natural. The intimidation was entirely in my head.

For any non-technical builder reading this: don't skip the fundamentals. The twenty minutes you spend learning basic Git workflow will pay dividends across every project.

The Speed of Progress

Let me put concrete numbers on this.

In roughly two hours of focused work, I went from a blank template to a fully-designed personal website with:

Custom dark theme with earthy color palette

Animated sections with scroll-triggered effects

Branded experience tiles matching company aesthetics

Integrated custom logo and headshot

Responsive design for mobile and desktop

Two complete pages (home and about) with consistent styling

A blog with this post you're reading

Two hours. And I don't know how to code.

This is the part that feels important to emphasize. The technology has reached a point where the bottleneck isn't capability—it's imagination and clear communication. If you can describe what you want, you can build it.

The Bottom Line

Building with AI isn't about replacing developers or learning to code. It's about having a new kind of creative partner—one that can translate your vision into reality at the speed of conversation.

The workflow that worked for me: start with references, use the right model for each task, iterate rapidly, and don't be afraid of the developer tools that seemed intimidating at first glance.

If you're a non-technical leader wondering whether you should invest time in learning these tools: yes. Absolutely yes. The hour you spend building something real will teach you more than any article (including this one).

Just start. The tools are waiting.