trueroot.ai — The evidence architecture behind your AI representation

Leave your AI narrative to chance and it drifts. Trueroot's method anchors it — correctly first, positively next.

Prompt trackers show you what AI already says about you. Trueroot works the other end: a method with a clear beginning and end that puts your evidence architecture in order. Your True Root^TM anchors who you are. Every other claim gets its own proof: why you can be trusted, why you are worth the price, why you are the expert. Get that right and the picture propagates from the inside out, instead of drifting.

One root, and proof for every claim.

Evidence architecture

An evidence architecture is the set of sources a language model draws on to describe you, put in order. At its centre is the True Root: the canonical, machine-correct entity page that says who you are. Around it are the pages that prove every other claim, such as your standards, your pricing and your methods. Each is linked from the root, and all of them tell the same story.

Dashboards show results after the fact. They don't solve the problem.

You can track prompts endlessly: how often you’re mentioned, in what wording, alongside whom. Useful for sure: it tells you something is going on.

Tracking is a measurement after the fact: the output. If you want to change those results though, you need to work at the causes of it.

So the solution is on the input side. You put the sources the models read in order: your evidence architecture. And the source you start from is the True Root^TM.

What you can’t see now

Whether you’re mentioned when someone asks about your category.
Whether you’re described correctly, or with outdated or invented facts.
Whether the picture drifts from what you’d say yourself.

Do you qualify? Start with the check.

An evidence architecture only works on a foundation that’s sound. So you check your site first, before an intake makes sense. You fill in what applies: yes, no, or don’t know.

Are your key pages readable without JavaScript?

The text is in the HTML itself, not only after scripts load.

Do your pages load fast?

A page is fully visible within a few seconds.

Are copy and legal on board?

You can have the texts adjusted and have the rights and the right facts to publish.

Schedule an intake →

Not a yes to everything, or a don’t-know here and there? That’s fine. We’ll take it into the intake.

Not one page. The evidence behind every claim, put in order.

Most organisations already have an about page. It just rarely says, in plain language, what the organisation is, what it belongs to and on what it should be judged. A visitor can fill in the blanks. A language model can’t: it has to piece the facts together itself, and fills the gaps with assumptions.

trueroot starts there, and rewrites that page into a True Root: the anchor for who you are. All existing content stays. The facts, the relationships and the definition are stated clearly, written so a human reads them comfortably and a model can take them over cleanly.

But one page cannot carry every claim. An about page says who you are. It does not prove that you can be trusted, that you are worth the price or that you are the expert. AI looks for different evidence for each of those questions. Research by Steven Wilson-Beales into 4,003 AI answers about news publishers showed it, and our own research with 25 Dutch brands in eight sectors shows the same.

That is the evidence architecture: the True Root as the anchor, and proof for every other claim, each linked from the root and telling the same story everywhere. No separate machine layer next to your real pages, but texts that serve both audiences at once.

From intake to a delivered evidence architecture.

A defined track, after a successful check. Short, focused, and with your own people at the table where it matters.

Intake. We determine the entity, its audiences and the claims that need proof: who you are, why you can be trusted, why you are worth the price, why you are the expert. This surfaces where that proof is now and where it is missing, and we discuss the third condition: rights, facts, copy and legal.

Working with your stakeholders. We don’t work on your pages solo. We align with copy and legal: copy guards the voice and brand language, legal guards what may factually be claimed. That way the evidence stays yours, not ours.

The True Root. We rewrite the about page into a True Root. Existing content stays, the definition comes first, the facts become independently readable, and the relationships with parent company, category and people are made explicit.

Evidence per claim. For every other claim we set out the proof it needs and where it should live: your standards and complaints procedure, a record of corrections, pricing that explains what is included, your methods with the experts named. We write or rewrite those pages with you and link each one from the True Root. Outside your site, we list the registers and profiles that must say the same.

Structured data and delivery. We capture the same facts in the machine-readable layer, so text and data broadcast the same thing. You get the pages and the structured data, ready to publish, with a clear explanation of what changed and why.

The evidence architecture is the starting point. Not the end point.

A one-off track sets the sources right. What the models do with them afterwards keeps moving. Optionally we keep an eye on that and build further.

Optional, ongoing

Monitoring

We monitor weekly whether AI still portrays you as your True Root and your evidence state it: does AI mention you, describe you correctly, and is the picture not drifting. If a gap appears, we escalate and adjust.

For an additional fee

Optional, on request

Additional tracks

If you want to go beyond your own sites, towards broader AI visibility and the external picture that feeds your representation, follow-up tracks are possible. We look at what makes sense case by case.

By arrangement

The analysis doesn’t come from a tool, but from someone who knows the craft from the inside.

trueroot.ai is run by Ramon Eijkemans, with more than 20 years of experience in enterprise SEO on complex, large-scale websites. In recent years he has focused intensively on the findability and representation of brands in language models, with his own research and a framework that is being picked up in the field.

His work has been featured in Search Engine Land and Search Engine Journal, and in the podcast The Search Session he discussed at length with Gianluca Fiorelli how to write content that language models can read and represent correctly.

That’s deliberately the heart of trueroot: you don’t build an evidence architecture with a generator. It takes someone who understands how models select content, how an entity is built up, and where the facts have to be right. The ambition is to expand the team with more seniors from the field, so trueroot can serve more organisations without compromising on the level of the analysis.