For ten years I designed interfaces that existed on a screen. A user saw them, touched them, navigated through them. My job was to make that experience clear, intuitive, and beautiful. That was it.

But something changed and it took me a while to notice. The interfaces I design now are not only seen by people. They are read by machines. Machines that extract meaning, build context, and generate responses based on what they understand from my content. And I discovered that the way I write code and content directly affects how those machines interpret me.

I stopped designing static interfaces. I started designing for machines that think.

GEO: the concept nobody taught you in design school

Generative Engine Optimization is the practice of optimizing your content for generative engines, not search engines. The difference is fundamental. A search engine puts you on a list. A generative engine puts you inside an answer. Or does not put you at all.

As a designer this fascinates me because it is a communication problem. We are designing for an audience we never considered: the machine that reads your content before the human sees it. If that machine does not understand you, the human never gets to read you.

I implemented GEO on my own site without knowing it was called that. I started writing more direct, more specific, more citable content. Then I added semantic structure with JSON-LD. Then I implemented llms.txt. Each step was a design iteration, not a marketing one.

Custom Knowledge Graph: your blog as a knowledge database

Most developers know schema.org. They add a BlogPosting with title, date, and author. They check the structured data box and move on. That is the minimum. It is like delivering a wireframe and calling it the final design.

What I built is different. Every post on my blog automatically generates a knowledge graph that connects the article with other related articles through relatedLink, identifies content topics as typed entities with about, and detects mentioned tools with mentions. All automatic. All based on the actual content.

That transforms my blog from a chronological list of articles into a semantic network that AI models can navigate. When ChatGPT processes my site it does not see fifty independent pages. It sees an interconnected knowledge system about a designer with fintech experience documenting his use of AI. That is the difference between being another page and being a source.

Sentiment Mapping: the tone of your code matters

This is the concept fewer people understand and the one with the most impact. LLMs do not just read what you say. They read how you say it. And they make decisions based on that.

When I write I implemented this on my site and the results changed an LLM interprets that as experiential evidence. It gives it weight. When someone writes this might work the model interprets that as speculation. It gives it less weight.

That means the sentiment of your content, the confidence with which you write, the specificity of your claims, directly affects whether a language model cites you or ignores you. It is not just technical semantics. It is tone. It is voice. It is feeling translated into citation probability.

As a designer this makes perfect sense. We always knew that the visual tone of an interface affects how the user trusts it. Now the textual tone of your content affects how an AI trusts you as a source.

What 90% of developers still do not understand

Most developers still think of SEO as the main game. Keywords, backlinks, load speed. All of that still matters but it is no longer enough. The game changed and most did not notice.

GEO is not an extension of SEO. It is a different paradigm. It requires thinking like an experience designer, not like a keyword optimizer. It requires understanding that your audience is no longer only human. And it requires accepting that how you structure your content determines whether AI models treat you as a knowledge source or as noise.

The developers who master these three concepts, GEO, custom Knowledge Graph, and Sentiment Mapping, will have a brutal advantage in the next two years. Not because they are better programmers but because they understand something most people ignore.

Code has feeling. And the machines are reading it.