I use AI tools to help with my website and other projects. They help me work through source files, draft changes, and explain unfamiliar parts of the code.

I still have to decide what I want and check what changed. A confident answer is easy to accept when I’m trying to move quickly.

Give it one clear job

A useful request names the page, the problem, and what should happen. “Make everything better” leaves too much room for assumptions.

For my portfolio, I asked for specific changes: separate creator socials from professional contacts, open the CV in a new tab, and keep logos readable in both themes.

Open the result

I’ve used Codex and Antigravity for guided project work. My part includes setting the scope, questioning claims, and reviewing the actual page.

A footer mark can disappear on a dark background. A screenshot can show the current public website while the text discusses an earlier internship. Those details need a visual or content check.

Keep the claims accurate

My resume separates basic familiarity with AI tools from tools I have used in project work. I do not want a logo strip to look like an expert rating.

The same rule applies to case studies. Personal projects, an internship contribution, and freelance annotation work need their own labels.

My finishing checklist

  • Does the change solve the task I described?
  • Do the links and controls work?
  • Is the page readable at a smaller width?
  • Are the screenshots and experience claims accurate?
  • What is still incomplete or uncertain?

Build and type checks help with technical issues. Browser review helps with the experience someone actually sees. I need both when the task involves implementation.

I’m still learning through this process. The goal is a result I can explain and stand behind, even when AI helped me build it.

Keep the request small enough to review

A focused revision gives me a clearer comparison between the problem and the result. If I ask for a footer adjustment, I can inspect that component and check its appearance in both themes. If the task changes content, I can compare the wording with my actual experience before accepting it.

The same discipline applies when the AI suggests a wider rewrite. I need a reason for the extra scope and a way to check that existing behavior still works. My role is to set the direction and review the outcome. AI assistance helps me make progress while learning, but the finished portfolio still needs to represent my own work accurately.