12% of U.S. adults are on GLP-1 medications and turning to AI for food guidance. We ran 120 queries across ChatGPT and Gemini — with and without GLP-1 user memory context — to find out which food brands AI recommends, and whether memory changes everything.
Research Scope
We ran 120 queries across ChatGPT and Gemini — half with a simulated GLP-1 user persona, half without. Here's what we found.
Magic Spoon leads with a 20% mention rate in the persona condition. 9 of 21 tracked brands — including many nutritionally aligned legacy CPG brands — have zero AI mentions in both conditions.
Magic Spoon surges from 4 cold mentions to 12 persona mentions (+200%) when AI 'remembers' the user is on Ozempic. Legacy brands like Cheerios (−25%) and Fiber One (−67%) lose ground as memory activates.
Gemini favors mainstream brands (Cheerios leads cold). ChatGPT surfaces DTC challengers (ALOHA, Catalina Crunch, :ratio). ALOHA is the only brand appearing consistently across both engines and both conditions.
Cleveland Clinic, EatingWell, and Trimrx.com (GLP-1 clinic) dominate citations. Only one brand domain appears meaningfully in AI citations — suggesting earned content authority is achievable but rare.
Brand awareness does not translate to AI visibility in the GLP-1 nutrition category. The brands winning AI recommendations aren't the most recognized — they have more AI-legible content, more cited sources in medical media, and a clearer GLP-1 narrative. As ChatGPT and Gemini expand persistent memory, this gap will compound. The window to establish AI authority in this category is open now.
Brand rankings across 60 queries per condition. Left: with GLP-1 persona context active. Right: cold user with no prior AI memory. The same 10 prompts, two very different recommendation sets.
% of 60 queries where AI cited this brand
Position-weighted score. 1st = 100, 2nd = 80, 3rd = 60... Higher = better placement
Average list position when mentioned. Lower = recommended earlier
Always recommended first when mentioned
Always recommended first when mentioned
These brands have zero AI mentions across 120 total queries. Despite relevant product portfolios, they are completely absent from GLP-1 food guidance conversations — invisible in the channel where GLP-1 consumers actively seek purchase guidance.
The same 10 prompts, the same two engines — but one condition includes a simulated GLP-1 user persona (female, 38, 3 months on Ozempic). The composition of recommendations shifts dramatically.
Disappears completely with persona context despite high-protein positioning
Stable across conditions — consistent AI authority
Engine-by-engine brand performance across both study conditions. The two platforms have fundamentally different GLP-1 food personalities — and they diverge further when memory is active.
“Mainstream nutritional authority — favors established brands and trusted food categories. Cheerios dominates.”
“DTC-forward and category-specific — surfaces challenger brands, niche DTC players, and functional food brands over mainstream CPG.”
“With memory, Gemini amplifies its existing preferences — Magic Spoon surges to 9 mentions while retaining Cheerios. More concentrated, less diverse.”
“With memory, ChatGPT becomes more DTC-specialist. Premier Protein enters. :ratio disappears. Recommendations are more granular and protein-focused.”
ALOHA is the only brand that appears consistently across both ChatGPT and Gemini in both study conditions — with a perfect 100.0 visibility score, always recommended first when mentioned. This suggests ALOHA has built genuine cross-platform AI authority in the GLP-1 nutrition space, not platform-specific visibility.
487 citations extracted across 120 queries (avg 4.1 per query). The sources AI trusts to inform food recommendations reveal the citation pathway brands need to build.
of U.S. adults have tried a GLP-1 medication
Up from 6% in May 2024 — adoption doubled in 18 months
KFF Health Tracking Poll — November 2025
of GLP-1 users report their entire household's diet has changed
FleishmanHillard, 'Understanding the GLP-1 Consumer' — October 2025
average decline in household grocery spending within 6 months of GLP-1 adoption
Numerator household panel of ~150,000 U.S. households
Hristakeva, Liaukonytė & Feler, Journal of Marketing Research — December 2025
are the only food categories showing spending increases among GLP-1 users
Hristakeva, Liaukonytė & Feler, Journal of Marketing Research — December 2025
Directional — 10 prompts per condition; below 50-prompt standard for full benchmark. All findings labeled accordingly.
This is an exploratory pilot study. The A/B design is methodologically novel; the memory effect direction is consistent across both engines. Brand-level findings are directional only. A full-scale GLP-1 benchmark with 80–150 prompts and 3+ platforms would yield publishable high-confidence findings.
ChatGPT and Gemini selected because they are the only major AI platforms with production memory features. Together they represent ~75% of U.S. AI market share. Perplexity excluded (no memory feature; citation-heavy search behavior differs from conversational recommendation).
Female, 38, 3 months on Ozempic (0.5–1mg), 14 lbs lost, protein-focused diet, nausea and constipation concerns, grab-and-go food preference, prior cereal affinity. Designed to simulate the memory state a returning GLP-1 user would have built in ChatGPT or Gemini after 3 months of food guidance conversations.
No system prompt. Each query treated as a new user with no prior relationship to the AI platform. Baseline for uncontextualized GLP-1 food recommendations.
e.g. “What are the best high-protein cereals I can eat while on Ozempic?”
e.g. “What's a good on-the-go snack bar for GLP-1 users that's high in fiber and protein?”
e.g. “I'm dealing with constipation from my GLP-1 medication. What high-fiber snacks do you recommend?”
e.g. “Give me a simple one-week high-protein meal plan for someone on a GLP-1 drug.”
Get a personalized analysis of how your brand performs across ChatGPT, Google AI, and Perplexity— plus actionable recommendations to improve your visibility.