• AI
  • Process
  • Portfolio

I started this portfolio one year ago. AI changed its direction twice

The story of this portfolio: AI as a thinking partner, then a coding partner, then a whole team.

AI and automation were an important part of this portfolio from the very beginning. But when I made my first commit, just over a year ago, I didn’t expect the direction that felt so unique to me back then to change twice in a year.

Why did it change? Here’s the story of how this portfolio came to be.

AI as a thinking partner and assistant

In autumn 2025, I used ChatGPT to brainstorm ideas for this portfolio, plan my Jira items and milestones, and keep track of the work. I was amazed at how well the Atlassian connector for ChatGPT worked. In my first Jira setup, I created all the tasks by hand, and AI wrote some of the descriptions.

In the second one, I let ChatGPT manage Jira for me. I could focus on the work and just ask “What’s the next step?” or say “Write this in a comment and close the task”.

At this stage, ChatGPT was also my assistant. It helped me set up my GitHub repos and start the portfolio page from a public Astro template. It feels strange that this first version is only a year old.

I also had to refresh my memory of Postman for API testing and of version control: branching, committing, pushing, merging, and deploying. I was glad that AI could help me learn and do these things on my own.

Back then, I saw the technical writer as a bridge between technology, product, and users. Writers built that bridge through their own work: they wrote, designed automations, and proposed a direction for the product. I wanted my portfolio to show as many types of documentation as possible. Documenting the process itself mattered to me too.

AI as a coding partner

In early spring 2026 came the first big shift in how I worked on the portfolio. I started using Antigravity.

Again, I was amazed at how easy and efficient it was to build with it. I rebuilt the whole portfolio and moved it further away from the template I had started with. It became more mine.

Then I discovered Codex. The work got easier and more efficient again, and faster too. I could work on my blog posts, my Jira items, the API documentation, and the portfolio page at the same time.

At this stage, the content of the portfolio didn’t change much. The way I worked on it did. AI took over more and more of the writing, the automation, and the version control. I removed one thing: the automation case study. With today’s AI and its connectors, it no longer made sense. That was the first change of direction: automation was no longer something my portfolio showed.

Around that time, an idea for another project started to take shape. I was struggling with motivation, so I started rewarding myself with points for everything I got done. With enough points, I could buy myself something, for example a LEGO set (you should know that I’m a LEGO maniac). After a couple of weeks, I replaced the points with stickers. A couple of weeks later, my family joined the quest. It was only a family game. For now.

AI as my team

Over the summer, I didn’t work much on the portfolio, while the world of AI kept moving fast. When I came back to it in October, another change of direction was inevitable.

I started using Claude Code. I still used Codex alongside it, but not much for coding.

I was lucky: in my version of Claude Code, the new projects and threads functionality worked out of the box (I’ll tell you one day how I lost it). This UX was so intuitive for me that I worked much faster. In a week, I built a prototype of my own brand, StickerStart, based on the family sticker system I mentioned above.

And then it hit me: the direction of my portfolio no longer made sense. I could see in my own work how quickly the technical writer’s job was changing, so the portfolio had to show it too.

Of the projects in the previous version, only the API documentation stayed. AI gave me the tools to create and manage a whole product, and I wanted to show these skills. So I decided that StickerStart would become a case study in my portfolio. It will show how I approach a brand and a product through their whole lifecycle.

While working on StickerStart, I used the Promo Video Skill, a Claude skill created by my husband, to make a video about the project. I thought: this skill is a great thing to document. So I also added classic user documentation, as I had planned from the beginning: the docs for the Promo Video Skill.

And what about AI as my team? It went from a thinking partner, through a coding partner, to a whole team of agents working for me. I think the story of my portfolio shows how fast AI evolves, and how people in different professions can put it to use.

Now I have a project orchestrator for each of my projects. When I need to work on the page, I set up a team of agents to work on solutions for me: for example, a UX designer, a graphic designer, a developer, and a tester. I give AI the direction, and it carries out my ideas. The world of technology no longer restricts me. It lets me thrive.

The technical writer is still a bridge between technology, product, and users. What changed is who does the work on the bridge.