Not long ago, a client looking for a branding studio in Dubai would search Google, open a few tabs, and build a shortlist by hand. Today, that same person might ask ChatGPT or Perplexity: “Who handles rebrands for fintech startups entering the Gulf?” The AI answers in seconds, with a name or two.
At Svyazi, a creative agency, we see this shift in our own inbound pipeline. Inquiry sources are tracked through analytics and aggregated in our CRM. Today, around 15 qualified inquiries a month come through AI-assisted discovery. For years, digital visibility was built around earning a place on page one. Now it also means earning a mention inside the answer itself.
Search is not dead, just no longer the whole journey
None of this means traditional search is dead. People still use it to check prices, read reviews, compare options, and verify what is current. The change is where the first impression begins. By the time a prospect reaches a website, the brand may already have been framed by an AI answer.
That makes discoverability a brand issue as much as a search issue. SEO helps a page get found and checked. But AI visibility also depends on positioning, content, PR, and reputation working in the same direction: what the company is known for, which industries it serves, what problems it solves, and what proof exists beyond its own website.
What AI models actually look for
Five kinds of signals seem to decide whether a brand gets recommended.
Relevance and positioning. AI tools are matching a brand to a specific situation, not a category. Being a “branding agency” answers almost nothing when a model has thousands of those to choose from. Being the agency a fintech founder calls before a rebrand is a real answer. That specificity has to live in public language, not inside a founder’s head: service pages named after the client’s situation, case studies titled by the problem solved, an “about” page that says exactly who the brand is for.
Expertise. What AI is weighing here is a demonstrated point of view. It gets built mostly through content published somewhere other than the brand’s own site: articles under a named author, commentary picked up by trade press, opinions credited to an actual specialist at the company.
Extractability. The skill here is genuinely new: how easily a model can lift a fact or an argument out of a page and reuse it. A strong point buried in the fourth paragraph of a long page is close to invisible to a tool skimming for an answer. Building this is about how content gets written. What helps:
- Clear headings
- Short sections built around one idea
- Direct summaries
- Question-and-answer formatting
Outcomes. “We deliver high-quality solutions” carries no evidence, and a model has no use for a sentence like that. What it is looking for instead is proof that the brand has actually done the thing it claims to do, and that proof needs to be visible publicly, not filed away in a pitch deck. What carries weight:
- Specific figures and dates
- Named clients
- Case studies with a clear problem, process, and outcome
- A described way of working
Credibility. This is the signal a brand cannot manufacture alone: something outside its own marketing that backs up the claim. Credibility that only the brand vouches for is not really credibility. It comes from press coverage, industry rankings, reviews, and listings on sites the brand does not control, sources a model can treat as independent of the brand’s own voice.
How to check whether AI can actually see a brand
A quick way to check where a brand stands: skip the obvious question. Asking an AI tool “What do you know about us?” only checks whether the model can produce a fact sheet, not whether a real customer would ever find the brand. A better test is the question a stranger would ask, without naming the company: “Who handles branding for hospitality groups in Riyadh?”
Then look closely at what comes back:
- Does the brand appear without being prompted?
- How does the AI describe what it does? Independent language, or a quiet echo of the brand’s own marketing copy.
- Who else gets named alongside it? That is the competitive set assigned to the brand.
- Where do competitors show up that this brand does not? Rankings, articles, and reviews the brand is missing from often explain why.
- Are the market, specialization, and format of work described correctly?
- What is the answer based on? Some tools will show their sources if asked, often the fastest way to spot an outdated page or a skewed listing.
Brands need to become easier to understand
Search rewarded being findable. AI is asking for more: can this brand explain itself clearly, and does anything outside its website back that up. A vague brand could still capture traffic if the page matched the right keywords. In AI answers, that vagueness becomes harder to hide. Clarity and outside proof are what get a brand into the answer now.
(Ilya Zmienko is founder of Svyazi, a creative agency working at the intersection of graphic design, communications and AI-assisted content)
