Most agencies selling AI search visibility have a problem they'd rather you not notice: their own websites aren't built the way they're telling you to build yours. Ask to see their structured data, their llms.txt, their entity markup — and you'll find a WordPress theme with a plugin or two.
This case study is different, because you can audit it yourself, right now, from your browser. The subject is the site you're reading — pfaffdigital.com — which we built as the reference implementation for our AI Search Visibility service. Everything we recommend to clients is running here first.
The Objective: Build for Two Audiences at Once
Every website now has two kinds of readers. Humans, who need design, speed, and clarity. And machines — search crawlers, AI assistants, and increasingly the AI agents that research and act on a person's behalf — which need structure, consistency, and explicit answers.
Most sites are built for the first audience and hope for the best with the second. We designed this site for both from the first commit, because the second audience increasingly decides whether the first one ever finds you: when someone asks ChatGPT or Perplexity who to hire, the answer is assembled from what machines could read and verify.
What's Under the Hood — and How to Check It
Structured data on every page. View the source of any page on this site and you'll find JSON-LD: Organization and Person entities site-wide, Service schema on every service page, FAQPage markup on every FAQ, Article markup on every post. AI systems don't have to infer what Pfaff Digital is, who runs it, or what it offers — it's stated in the format they parse natively.
Machine-readable site summaries. Visit /llms.txt and you'll find a clean, plain-text summary of the business — who we are, what we offer, how to reach us — in the emerging standard AI crawlers check. A fuller version lives at /llms-full.txt, and a structured plugin manifest at /.well-known/ai-plugin.json. Three formats, because different AI systems read different things, and we'd rather be legible in all of them.
An open door for AI crawlers. Our robots.txt explicitly welcomes GPTBot, ChatGPT-User, Claude-Web, Google-Extended, and the rest of the AI crawler population. Many sites block these bots by default and then wonder why AI engines never recommend them. Being cited starts with being readable.
Agent-ready interfaces. Beyond being readable, parts of this site are operable by AI agents — we run a working WebMCP demonstration that lets an AI assistant interact with the site directly. When your customers start sending agents to research and transact on their behalf (they've already started), the sites those agents can actually use will win.
Performance as an AI signal. The site is statically generated and served from Cloudflare's edge network, with optimized images, preloaded fonts, and no heavyweight client-side frameworks doing work the server already did. Speed isn't only a human courtesy — retrieval systems deprioritize slow sources.
Entity consistency. The name, location, services, and story of Pfaff Digital read identically in the page copy, the structured data, the llms.txt files, and our external profiles. Consistency is one of the strongest trust signals an AI system has — conflicting information quietly disqualifies a business from being recommended.
The Design Constraint Nobody Talks About
Here's what makes this genuinely hard, and why we treat it as an engineering discipline rather than a checklist: none of the machine-facing work is allowed to degrade the human experience. The structured data is invisible. The llms.txt costs nothing. The semantic markup is the accessible markup. The performance work serves both audiences at once.
A site built only for AI readers would be a spec sheet. A site built only for humans is increasingly invisible. The craft is in building one artifact that serves both — which is a design-and-engineering problem, and exactly the intersection where we work.
Verify, Then Talk to Us
We built this site to be the answer to a fair question: "If you're so good at AI visibility, show me." So — check the source. Open /llms.txt. Run any page through a structured-data validator. Ask an AI assistant what Pfaff Digital does and see what comes back.
Then, if you want the same engineering applied to your business, start where every engagement starts: the free AI Visibility Report. We'll show you exactly how AI systems see your site today — and what it would take to make you the recommendation.