Sara & Nik’s Podcast:
Antonis Dimitriou

About this conversation

Antonis Dimitriou, an SEO expert with 6 years in the industry, joins Sara & Nik to discuss everything AI-related, from optimizing content for AI models to how you can manipulate LLMs and whether AI Mode is ready for users.

His experience in B2B content strategy and SaaS serve as the backbone of his SEO strategy which, in 2026, revolves around optimizing the content for AI visibility.

Why Optimizing for AI is Good SEO Practice

Sara’s question sets the topic of the discussion right from the start: how does Antonis talk to clients who distrust AI?

Antonis has a clear-cut strategy for that: base everything on measurable facts. So, his first line of attack is to show clients verifiable statistics that compare AI search results with traditional search engines. The data is clear – people use AI differently in their search queries compared to standard search engines.

However, this doesn’t change the fact that SEO, the underlying foundation, has remained the same. In Antonis’s view, with or without AI, SEO works identically. For this reason, he doesn’t bother optimizing the content for each AI tool in particular; instead, he optimizes across platforms.

The data supports this as well, as Antonis explains, given that 90% of AI-targeted SEO is “simply good SEO.” The process works the same:

Create your website → Build social profiles → Produce on and off-site content, etc.

It’s all the standard SEO steps that will help a brand show up in the training data.

The rest of 10% is just basic product details.

The Limits of LLMs

Nik brings up the idea of being active on forums, review sites, and social media platforms to increase the brand’s visibility, which isn’t new. The difference, as Antonis notices, is that it’s even more important today, compared to the past. The main differentiating factor being AI.

“Right now I think it's more tangible because you actually see that it is used.”

He gives the examples of GPT and Perplexity pulling results from Reddit in real time during their targeted searches and they do it faster and more effectively than traditional search engines.

The problem, as Sara observes, is that the industry hasn’t adjusted to the changes yet. Auditing a client’s Youtube channel or Reddit activity was disconnected from the actual SEO work in the past, which is no longer the case today. The AI has changed the SEO landscape, making it almost mandatory to do SEO across the board, including Youtube videos.

"I think this should be pretty much standard practice if your client has a channel".

But then the quantity-over-quality dilemma comes up, with Nik sharing a test he did within his own agency: only a handful of listicles, which rated his agency at the top, were enough to have his brand mentioned by GPT. Is this evidence that GPT, and LLMs in general, prioritize quantity over quality?

Antonis believes that the answer depends very much on the query but that, in general, quality is more important, especially for specific niches like “the best agency for X budget/niche”. For this reason, he believes that AI companies will keep improving and add stronger quality signals over time.

Sara remains unconvinced. In her view:

"I don't think LLMs are ready for the amount of spam that's coming their way"

The problem, she thinks, is that Google has vast experience with spam content, but LLMs do not; they vastly rely on user feedback, which explains why they keep shifting their answers and “change their output every time.”

This underlying problem explains why LLMs still hallucinate today, making Google look more reliable and trustworthy.

About Budget Spending and the AI Mode

The discussion adopts a more practical angle now, with Nik offering a clear-cut scenario: A B2B SaaS client with a monthly budget of $5,000 and virtually no SEO.

Antonis lays out the basics first, starting with the foundational links and technical setup and ending with the canonicals. Then comes the bottom-of-funnel content, which moves into top-of-funnel topics designed “to boost authority to the money makers which are at the bottom of the funnel pages.”

Finally, he would repurpose every piece of content for social media platforms like Reddit which, he admits, many clients are hesitant about. The reason is the old mentality which translated SEO into basic linking and producing keyword-optimized content, with new channels like social media management requiring news skills and budgets.

When it comes to the AI Mode, Sara is very much unconvinced. The AI Overview section gives her the right answers quickly, which is enough most of the time. When it isn’t, she prefers to go to a trusted link instead.

Antonis agrees and describes AI Mode as unpredictable, with the same query often triggering different background searches. Nik follows the same line of thinking, also believing that the AI Overviews section is better overall than the AI mode.

That being said, AI Overviews isn’t the golden standard yet. To this point, Sara shares a piece of knowledge she got from scrolling through some Department of Justice documents: the AI Overviews engine uses a faster, but more limited search system, which often results in lower-quality results.

Antonis, who is familiar with Sara’s newsletter on the topic, remains optimistic. The way he sees it, this system is the result of prioritizing speed over the older models that were slower. With that out of the way, he expects improvements in the months to come, because:

“Google wants to be an AI first answer engine”.