How we set sites up for search engines and AI
The concrete, unglamorous technical work behind visibility: structured data, a sitemap, an llms.txt file, and tools that let AI agents read a site directly.
Most of what makes a site visible, to search engines and increasingly to AI systems, is invisible to a visitor. It's a handful of technical decisions made at build time rather than a marketing campaign layered on afterward. This is what we set up by default.
Structured data on every page
Every page carries schema markup describing what it actually is: organization details, article metadata on guides, breadcrumb trails so a search engine or an AI model understands where a page sits in the site, not just that it exists. It's a small, plain-language description written for machines, sitting quietly in the page source.
A sitemap that stays accurate on its own
New pages, whether that's a portfolio piece or a knowledge base guide, get added to the sitemap automatically because it's generated from the actual content rather than maintained as a separate list. A sitemap that drifts out of date is worse than no sitemap, since it tells search engines about pages that no longer exist or misses new ones entirely.
An llms.txt file
Alongside the usual robots.txt and sitemap, we publish an llms.txt: a short, structured summary of who the business is, what it does, and which pages matter most, written specifically for AI systems reading the site. It's the same idea as a sitemap, aimed at a different kind of reader.
Tools an AI agent can call directly
Where the browser supports it, we expose a small set of tools an AI agent can query directly: company information, services offered, a link to the portfolio, a way to open a contact request. Instead of an agent scraping a page and guessing at what matters, it can ask for the answer directly. This is an early standard and still evolving, but the direction is clear: sites that make themselves legible to agents, not just to human visitors, get read more accurately.
Why this is a default, not an upsell
None of this is exotic engineering. It's the same instinct as writing a clean sitemap or compressing an image: remove friction between your content and whatever is trying to read it, human or otherwise. We build it into every site from the start rather than treating it as a separate line item, because retrofitting structured data and metadata onto an existing site is far more work than including it from day one.
See what SEO actually means and what GEO means for the fuller picture of what this technical work is actually for.
Related guides
- What GEO means, and why ranking isn't the whole game anymore
Generative Engine Optimization explained: how AI answer engines decide what to cite, and why it's a different job than ranking on a search results page.
- What SEO actually means
A plain explanation of search engine optimization, what actually moves rankings, and why most of it is just doing the technical basics right.