How I Used AI Agents to Build This Portfolio Faster
I did not build this portfolio because I wanted another personal website. I built it because I needed a better way to explain myself.
A resume can say “Senior WordPress Developer” or “Full-Stack Engineer”, but that does not really explain the full picture. In my case, I have worked with WordPress, PHP, Laravel, Alpine.js, Tailwind, Node.js, NestJS, APIs, automation, reporting systems, warehouse operations, audits, and team leadership. That is a lot to compress into one page without making it look like a keyword wall.
So I wanted my portfolio to work more like a recruiter filter.
If someone is hiring for WordPress, they should see my WordPress proof first. If they care about backend systems, they should see the projects and experience that match that. If they are looking for someone technical but also practical, the site should make that obvious without asking them to read everything.
That was the idea.
The interesting part is that I did not build it alone in the traditional way. I used AI agents as part of the process (the trend).
Not to replace the work. To move through the work faster.
The starting point
I already knew the direction I wanted.
One page. Strong visual identity. No generic “About Me, Skills, Projects, Contact” layout. I wanted something more interactive, with movement, role selection, and a hero animation that connected to the rest of the page.
The site became a WordPress theme using PHP, Tailwind, and Alpine.js. No React, no page builder, no ACF, no heavy admin system. The idea was to keep the code small enough that I could control it, but flexible enough that the content could change based on the role selected.
The main interaction is simple:
A recruiter chooses what they are hiring for.
The page changes the evidence.
That means the same portfolio can speak as a WordPress Engineer, PHP Full-Stack Developer, Backend / Platform Developer, DevOps / Cloud profile, or Technical Lead.
That decision was not generated by AI. That was the product idea.
AI helped me explore how to shape it.
How I used AI during the process
I used several AI tools throughout the project, each for a different purpose.
For design exploration, I primarily used ChatGPT. I would describe the type of portfolio I wanted, share examples of sites I liked, explain what felt generic, and ask for alternative layouts, section ideas, and interaction concepts. It was useful for quickly generating directions and helping me avoid staring at a blank page.
For visual concepts and mockups, I used AI image generation tools and Claude to experiment with different styles, hero sections, and layouts before writing code. Most of the concepts never made it into the final site, but they helped me figure out what felt right and what felt overdesigned.
For development, I used AI as a coding assistant (Cursor, GPT and Grok). I asked it to help structure the WordPress theme, organize configuration data, connect PHP data to Alpine.js state, build role-based filtering, and review implementation approaches. It was especially useful for creating first drafts of functions, exploring different patterns, and identifying edge cases I might have missed.
For the animated hero section, I used AI to brainstorm approaches for handling frame-based animations, canvas rendering, scroll interactions, and performance considerations. It helped me compare different implementation options before settling on the final approach.
For content, I used AI to rewrite sections, improve readability, and generate alternate versions of copy. Sometimes it helped me find a clearer way to explain something. Other times it produced text that sounded too corporate or too generic, so I rewrote it myself.
For reviews, I often used AI as a second set of eyes. I would paste code, layouts, or content and ask for feedback, potential issues, accessibility concerns, or simplification opportunities.
The biggest advantage was speed.
Instead of spending hours exploring a single direction, I could generate multiple options, compare them, reject most of them, combine the best parts, and keep moving.
But I did not blindly accept everything.
AI can generate code that looks correct while introducing unnecessary complexity. It can suggest features that sound impressive but do not improve the user experience. It can write copy that sounds polished but does not sound like me.
So every suggestion still had to pass the same questions:
Does this fit the architecture?
Does this make the site easier to maintain?
Does this improve the experience?
Does this support the story I am trying to tell?
The agents were fast. I had to be the filter.
One thing I learned quickly is that different AI tools are good at different things.
Some were better at design ideas.
Some were better at code generation.
Some were better at reviewing existing work.
Some were better at writing.
The challenge was not getting ideas. The challenge was deciding which ideas deserved to stay.
What AI was not good at
AI was not good at knowing what mattered most.
It could generate endless ideas, but it could not decide which ones aligned with my goals.
That responsibility stayed with me.
If I followed every suggestion, the portfolio would have become larger, more complicated, and less focused.
Instead, I kept bringing everything back to the same objective:
Show what I can do.
Show proof of experience.
Make it easy to understand.
Make it feel like me.
Another problem is that AI does not automatically know your real story. It can write a good-sounding version of you, but sometimes that version is not accurate. In one mockup, the copy said I grew up in Chile. That was close to part of my story, but not right. I am Venezuelan. I lived in Chile for eight years. Now I am in Indiana.
That difference matters.
The portfolio is supposed to represent me, not a polished fictional developer.
The real value was not automation
The real value was compression.
AI compressed the distance between idea and prototype.
It helped me move from “maybe my portfolio should be different” to a real direction. From direction to mockup. From mockup to code. From code to refinements. From refinements to sections that actually represent my experience.
But the important decisions still needed experience.
I knew why WordPress made sense for this portfolio.
I knew why Alpine.js was enough.
I knew why I did not want React for this version.
I knew why the site needed to be role-based.
I knew which parts of my experience were strongest for recruiters.
AI accelerated the work, but it did not replace the judgment.
Final thought
This portfolio was built faster because of AI, but it still had to become mine.
That is the part I care about.
The tools helped me move quickly, but the final result still depends on taste, experience, and knowing what story I am trying to tell.
For me, that is the future of working with AI.
Not replacing the person behind the work.
Making that person faster, sharper, and maybe a little less stuck.