ABOUT DEJAN
DEJAN is an Australian AI SEO agency with a global client base. We work out why AI systems recommend one brand over another, then make yours the one they name.
We started in 2008 as a search agency, back when winning meant a blue link near the top of a page. Over more than a thousand campaigns we built a habit that still defines us: test everything, measure what happened, and never accept a tactic we can’t explain.
In 2013 our founder published “Conversations With Google,” arguing that search would shift from a list of links to a dialogue that answers you directly. That prediction has arrived. People now ask assistants and AI search for a recommendation and act on the reply, so we rebuilt the agency around a single question: when an AI answers, is your brand in the answer?
Search has become something a model generates rather than a page it ranks. Being on page one no longer guarantees you are in the answer, because the model reads a handful of sources and decides, in a single pass, which brand to cite. That decision is what we optimize for.
We don’t think this needs a new acronym. It is still SEO, and the same people who earned visibility in search are the ones who will earn it inside AI answers. What has changed is the toolkit, and that is where we’ve gone deep.
Our founder trained a language model from scratch, from raw noise up, so we understand these systems from the inside rather than from the outside.
We probe what a model actually believes about a brand and steer it deliberately, using methods borrowed from mechanistic interpretability.
When the software to measure AI visibility didn’t exist, we wrote it. Our research runs on instruments we own end to end.
We are practitioners, not commentators. These are our own systems, used on live client work every day:
Tracks how often and how prominently a brand appears in answers from Google, ChatGPT, and other AI systems over time.
Measures how volatile AI answers are for a topic, so you know how settled or contested your visibility really is.
Reads a model token by token to reveal what it is likely to say about a brand, and where it is uncertain.
Turns model probing into a clear probability: how strongly a model associates your brand with an entity.
Reveals which domains and pages AI systems retrieve and cite for your topics, and where the gaps are.
Rewrites a page, round by round, until an AI ranker prefers it as the source worth quoting.
DEJAN is led by Dan Petrovic, one of the most recognised names in AI SEO worldwide and a double Google award winner. He trained a language model from scratch to understand how these systems choose their words, and shares that work constantly, on stage at Ahrefs Evolve, Semrush Spotlight, and SEO Week, and in near-daily writing.
Book a call with our senior strategy team to talk through your project.
