What I’d actually change on a site for AI search

Before adding anything to a site for AI search, I’d look at what its pages actually offer. An accurate answer can still be interchangeable with hundreds of others. A worked example, an honest trade-off or a useful comparison gives someone a reason to read yours. That’s where I’d start with answer engine optimisation, or AEO.

Google’s guidance supports that starting point. Its generative AI search features build on existing Search systems, and its advice still emphasises useful, distinctive content and sound SEO fundamentals. For developers, crawlability, accessible text and clear navigation remain part of the job. Making the page easy to reach matters alongside giving it something worthwhile to say.

On a portfolio, I’d put the problem and outcome near the top of each case study. Then explain the constraint that shaped the work: what had to stay compatible, why a particular approach made sense, or where the result fell short. Those details let a reader assess the decision. I’d add a diagram when it makes that decision easier to understand.

I’d also be specific about whose advice I’m following. Google says it ignores llms.txt for Search and doesn’t require pages to be split into tiny chunks. That helps me decide where to spend time on Google visibility; other services need their own checks. I’d put the effort into accurate, useful pages before adding machinery for a supposed ranking benefit. None of it guarantees a place in an AI answer.