This website, building my portfolio as a working product

The short version

The problem

A template portfolio would not show how I think, and would not sound like me.

What I did

Built this site from scratch with AI as the building partner, keeping tone, structure and judgement mine.

What it was worth

A live product I keep refining while using it for my own job search.

The call

Treated it as a working product, not a brochure. It changes because I use it.

I did not want a portfolio that felt like a cleaned-up CV or a template with my name dropped into it.

I wanted a working product that could show how I think, what I have built, and how I like to shape the awkward space between idea, logic, wording, and delivery.

That mattered for two reasons. First, the site needed to sound like me rather than like generic portfolio language. Second, I wanted a place where I could keep my product sense active by building, testing, fixing, and refining something real.

I built the site from scratch with AI as a fast building partner. Claude and Codex wrote essentially all the code, while I stayed responsible for tone, scope, hierarchy, and what deserved to stay.

The practical loop was simple.

• write or rewrite a section
• see it in context quickly
• click through the page properly
• notice what felt off
• refine the wording, spacing, or logic

That made the work faster, but it did not make the judgement automatic. AI helped me get rough ideas into the browser early. The useful part was being able to react to something real instead of discussing an abstract outline for too long.

The site became a small system rather than a single page.

• a homepage that draws out one change in full, then compresses four more
• a work page listing every example with what it was worth written next to it
• longer case studies for the two pieces of work that earn the space
• a method page, an about page and an experience page carrying the longer story
• a content structure that makes it easy to keep adding examples without rebuilding the whole site each time

The pages had to work together. The examples needed to support the claims on the homepage. The about page needed to deepen the story without turning into corporate profile copy.

Most of the useful work sat in the refinement rather than the first draft.

I kept adjusting the pace of the pages, the wording of headings, the order of examples, and how much detail each section should carry before it became too much. Some changes were visual. Some were structural. A surprising amount was just about making the tone feel more human and less rehearsed.

I also used the site as a place to keep smoothing rough edges, fixing bugs, tightening copy, improving navigation, checking mobile layouts, and removing little pieces of friction that would quietly make the experience worse.

This is one of the reasons I think building matters for product roles. Once you click through something properly, you notice the unclear transitions, the missing states, the parts that technically work but still feel wrong.

1. April 2026
   Reshaped the homepage so the prototype approach and proof points came through earlier and more clearly.
2. April 2026
   Refined examples copy and ordering so the page read more like practical work and less like a project gallery.
3. April 2026
   Adjusted wording across the site to better match my own voice and remove portfolio clichés.
4. April 2026
   Smoothed navigation, spacing, and mobile behaviour so the site felt calmer and easier to read.
5. April 2026
   Added this example to document the site itself as part of the same practical building practice.
6. August 2026
   Rebuilt the whole site on a new dark design, with the work list restructured around what each change was worth.
7. August 2026
   Swapped the palette to light and restructured again. Own-time builds now lead the page, and the career work is a compact list of expandable rows.

This site reminded me that AI is fast, but judgement is slower.

It is easy to generate sections, layouts, and variations quickly. The harder part is deciding what actually sounds true, what belongs on the page, what is trying too hard, and what still needs another pass. That is where the work still feels very recognisably like product work.

It also reinforced something I already believed. Building real things helps me stay sharper as a Product Owner. It keeps my eye on detail, makes trade-offs visible, and gives me a more grounded sense of what is awkward, what is clear, and what still needs simplifying before other people have to carry it forward.

I see this kind of work as a practical way to make ideas testable earlier and have better conversations with the people who will take it further.