My firm just closed out a study of 558 AI-generated answers about peptide clinics and weight-loss providers, and the headline number is not the one that should worry a brand. Defy Medical came out on top in 5WPR's Peptides AI Visibility Index 2026, named in 12.4% of answers. But that lead sits inside the same statistical range as Ro, Hims, BPC-157 and a non-brand entity the study calls Overview. Four major AI engines, asked the same 48 questions, could not agree on a single leader. That is the real finding, and it is the one I want brand and reputation teams to sit with.
The Google Playbook Does Not Transfer
I have spent decades telling clients that rank is reputation. If you are not on page one, you do not exist to the buyer. That rule does not disappear with AI search, but it stops being sufficient. ChatGPT, Gemini, Perplexity and Claude each run their own retrieval tool, on their own schedule, against their own index. In this study, ChatGPT named Defy Medical in 26 of 128 answers. Gemini, asked the identical questions, named a different entity, Overview, in 56 of 144 answers, and the other three engines never named it once. A brand can be strong on one engine and functionally invisible on another, at the same time, in the same category.
I saw the same pattern play out in a different category when we ran the numbers on robo-advisor AI visibility: Betterment led on mentions, Fidelity led on citations, and neither engine treated the two firms the same way. The specific names change by category. The mechanism does not.
The Number That Should Actually Worry You
Appearing in an AI answer and being cited as its source are two different things, and the gap between them is where I think the next reputation crises in health and wellness will start. Among the answers in this study that carried retrieval evidence, the FDA's own site was cited 18.9% of the time, the highest rate measured. The best-performing commercial brand, Fountain Life, reached 5.9% on its own domain. Several visible brands, including Zepbound, Saxenda and Victoza, were cited from their own sites in exactly 0% of sourced answers. Meanwhile a third-party publisher, formblends.com, was cited in nearly a third of sourced answers, a higher rate than any brand in the category managed on its own turf.
Read that again: a brand can be named by name in an AI answer, repeatedly, while a site it does not control and cannot edit is doing the actual explaining. That is not a visibility gap. That is a loss of narrative control, and it is exactly the kind of exposure my crisis and reputation teams get called in on after the fact. I wrote about this same split, appearance without citation, when a wine brand showed up in 36.9% of AI answers but was cited from its own site in exactly 0% of them. Health and wellness brands are walking into the identical trap, with higher stakes.
Why This Category Is a Preview, Not an Outlier
Peptides and weight-loss compounds sit at the intersection of consumer demand, regulatory scrutiny and a crowded field of clinics, drugmakers and telehealth brands, which is exactly why the fragmentation shows up so clearly here. But nothing about the mechanism is unique to this category. Any brand in a regulated or heavily researched space, health, financial services, legal, will show the same pattern once someone measures it: engines disagreeing with each other, and third-party sites out-citing the brand on its own name.
What I'm Telling Clients to Do This Quarter
Three moves, in order. First, stop measuring "AI visibility" as one blended number. Measure it engine by engine, the way this study did, because a strong average can hide a complete absence on the one engine your buyers actually use. Second, build owned pages that directly answer the specific questions consumers ask before they choose, in plain language, so an engine has something of yours to cite instead of only a forum thread or a third-party aggregator. Third, track where a brand is named in the answer, not just whether it is named. Position 3 and position 12 are not the same outcome, even when both count as an "appearance."
None of this is theoretical. It is the same discipline I have applied to every reputation account I have run for over two decades, applied to a new set of engines that do not read a press release the way a journalist does.
CONCLUSION
AI search is not replacing the reputation fight, it is fragmenting it across four engines with four different sets of sources, and the peptide and weight-loss category just showed exactly what that looks like in the data.