Peec AI, the AI search analytics platform, introduced the analysis crew it has assembled to reply the query each model is now asking: why do ChatGPT, Perplexity, and Google’s AI Mode suggest some corporations over others – and the way can a model earn its place in these solutions?
Former enterprise Search engine optimization leaders, an explainable-AI researcher, and the engineers behind among the market’s earliest instruments are reverse-engineering AI search, feeding what they be taught straight right into a product constructed to assist shoppers earn visibility there.
As AI reply engines change the previous listing of blue hyperlinks, the principles for which manufacturers get really useful now sit inside fashions with no revealed rulebook – and nobody outdoors these corporations is aware of precisely how the alternatives are made. Peec AI’s response: rent the individuals who work these guidelines out by experiment, pairing twenty years of enterprise Search engine optimization with hands-on reverse-engineering of reply engines and machine-learning analysis into how fashions resolve.
That analysis flows straight into the product. Peec AI tracks how manufacturers seem throughout the most important reply engines – measuring their Generative Share of Voice in opposition to rivals, sentiment, and the sources every engine cites. These indicators turn out to be a prioritized set of suggestions that present groups improve their visibility in AI engines like google.
Decoding the black field
A lot of that work is led by Metehan Yeşilyurt and Tomek Rudzki. Yeşilyurt reverse-engineers the algorithmic habits of AI search and reply engines with out touching the underlying fashions’ supply code. As a substitute, he runs managed technical experiments, analyzes API community site visitors, and reads client-side configurations to deduce how platforms resolve which net content material to floor and cite.
Yeşilyurt’s findings, which he publishes brazenly on his personal weblog, have helped form the rising fields of Generative Engine Optimization (GEO) and Reply Engine Optimization (AEO) for manufacturers, in-house groups, and advertising businesses. By analyzing ChatGPT’s net interface and community site visitors, he decoded how the assistant decides which sources earn an inline quotation, surfacing proof that the habits displays a Reciprocal Rank Fusion method that aggregates and scores content material throughout a number of subqueries. In late 2025 he revealed an in depth breakdown of Perplexity AI, documenting greater than 59 distinct rating elements and patterns, amongst them the platform’s reliance on a shortlist of top-tier sources, semantic relevance to the question, content material freshness, and speedy user-engagement indicators. He has utilized the identical technique to Google Uncover and Google’s AI Mode, mapping the multi-stage content material pipelines and freshness “buckets” that seem to control which pages floor.
Tomek Rudzki has been decoding generative engines like google since earlier than the sector even had a reputation. Previous to becoming a member of Peec AI, Rudzki was a co-founder of ZipTie.dev – the primary industrial instrument (early 2024) devoted to monitoring model visibility in AI engines like google. Notably, the instrument tracked Google AI Overviews from its earliest days beneath the “SGE” banner, again when it was restricted to logged-in customers inside Google Labs. On this position, Tomek was instantly accountable for designing and constructing the module that optimized content material for giant language fashions, giving manufacturers a tangible option to affect whether or not and the way AI programs cited their sources. His duties included constructing an LLM-answer sentiment analyzer and a immediate generator.
That very same hands-on understanding now powers Peec’s AI-visibility suggestions – so a model sees not simply the way it’s performing throughout AI reply engines, however the particular change almost definitely to develop its visibility.
That intuition runs deep. Rudzki’s engineering thesis at Politechnika Opolska was titled Empirical Evaluation of the Google Search Engine Algorithm, and for his 2019 computer-science grasp’s he constructed software program that improved rankings in conventional engines like google utilizing Search engine optimization crawl information, server logs, and Google Search Console indicators. He has spoken at business conferences together with BrightonSEO and SMX.
Additionally Learn: AiThority Interview with Matej Bukovinski, Chief Expertise Officer at Nutrient
Constructed on deep Search engine optimization foundations
That experimental work is grounded by individuals who ran Google natural search at scale lengthy earlier than LLMs entered the image.
Malte Landwehr brings greater than 20 years of enterprise Search engine optimization and SaaS product management. He beforehand served as VP of Search engine optimization at idealo, one of many world’s largest price-comparison platforms, the place he spent 5 years main natural development and doubled Google natural site visitors, and as VP of Product at Searchmetrics, a globally acknowledged enterprise Search engine optimization software program suite. He left a senior company position to assist construct Peec AI – the product he had concluded the market was lacking.
David Konitzny brings a technical-Search engine optimization and international-search perspective formed on either side of the desk – company and in-house. He began out as a technical Search engine optimization at Peak Ace, one in all Germany’s most famous businesses, earlier than turning into Head of Natural Search at Raisin, a German fintech unicorn, the place he constructed multilingual, multi-regional methods throughout worldwide markets. When he returned to company work at KKP, that in-house expertise paid off: he already understood the ache factors dealing with insurance coverage and banking shoppers, and resolve them. KKP can be the place AI search turned his focus – and the method he developed there’s the one he brings to Peec: not simply optimizing a consumer’s area, however educating the entire advertising crew and giving bigger organizations a transparent roadmap to comply with. He pairs that technical depth with industrial intuition, and as we speak makes use of browser developer instruments and different hands-on strategies to learn the indicators that reveal how LLM programs behave beneath the hood.
From analysis to actionable suggestions
Translating these discoveries into working software program falls largely to Dr. Melissa Fasol, one of many core builders on the Peec AI platform. She earned a PhD in computational neuroscience on the College of Edinburgh, specializing in Explainable AI, sign processing, and machine studying – work targeted on precisely the issue AI search presents: making the choices of complicated, opaque programs clear and reliable.
Earlier than Peec AI, she based Tulia AI a startup that helps manufacturers perceive why LLMs suggest sure corporations over others – work for which she was shortlisted for the Investec Early Stage Entrepreneur of the Yr Award. That method aligns intently with Peec’s mission, and at Peec she helps flip the crew’s collective analysis into the platform’s Actions module – ranked, opportunity-scored suggestions on what a model ought to do subsequent to enhance its AI search presence.
“Anybody can measure model visibility. The onerous half is giving suggestions that really transfer the needle and could be scaled in a sustainable approach,” mentioned Malte Landwehr, Chief Product Officer of Peec AI. “Many techniques that at the moment work properly for short-term positive factors in AI visibility will cease working in a yr. Some even put your complete Search engine optimization visibility in danger! To keep away from that, you want individuals with deep Search engine optimization expertise who’re additionally in a position to deconstruct how LLMs work.”
