AI Recommendation Research · August 2026
A 600-response audit showing how brand visibility changes by buyer situation, platform and category boundary.
Functional beverages do not behave like one coherent market inside AI recommendations. Recess leads overall, but retail questions produce a different competitive field from calm, social and energy questions. The platforms also disagree sharply about which brands belong - and sometimes place brands inside the category that do not claim to be functional at all.
Recess leads on every platform, but most brands depend heavily on one or two systems. Counts below are response appearances, not raw name mentions.
| Rank | Brand | Overall / 600 | Claude / 150 | ChatGPT / 150 | Gemini / 150 | Perplexity / 150 |
|---|---|---|---|---|---|---|
| 1 | Recess | 189 | 90 | 29 | 45 | 25 |
| 2 | Kin Euphorics | 148 | 68 | 23 | 35 | 22 |
| 3 | OLIPOP | 112 | 48 | 21 | 34 | 9 |
| 4 | Liquid I.V. | 96 | 50 | 17 | 23 | 6 |
| 5 | Hiyo | 91 | 25 | 7 | 38 | 21 |
| 6 | Poppi | 84 | 42 | 7 | 23 | 12 |
| 7 | Celsius | 78 | 35 | 8 | 22 | 13 |
| 7 | Moment | 78 | 35 | 12 | 15 | 16 |
| 9 | Four Sigmatic | 73 | 34 | 20 | 13 | 6 |
| 10 | Nuun | 59 | 40 | 14 | 4 | 1 |
| 11 | De Soi | 53 | 12 | 1 | 23 | 17 |
| 12 | Ghia† | 51 | 24 | 6 | 13 | 8 |
| 13 | Odyssey Elixir | 50 | 12 | 0 | 31 | 7 |
| 14 | GT's Living Foods | 49 | 22 | 10 | 17 | 0 |
| 15 | Curious Elixirs | 48 | 18 | 1 | 25 | 4 |
| 15 | LMNT | 48 | 34 | 7 | 5 | 2 |
† Category crossover. Ghia is retained because AI recommended it in these functional-beverage situations, not because Ghia describes itself as a functional beverage. Red figures indicate zero or one appearance on a platform. Rank 15 contains a tie.
The dominant brand
Recess appeared in 189 of 600 responses - 41 more than Kin Euphorics - and ranked first on all four platforms.
These are established brands in the overall ranking. Their weakness is concentrated on a particular platform rather than across the entire category.
| Brand | Strongest platform | Responses | Weakest platform | Responses |
|---|---|---|---|---|
| Nuun | Claude | 40 | Perplexity | 1 |
| De Soi | Gemini | 23 | ChatGPT | 1 |
| Odyssey Elixir | Gemini | 31 | ChatGPT | 0 |
| GT's Living Foods | Claude | 22 | Perplexity | 0 |
| Curious Elixirs | Gemini | 25 | ChatGPT | 1 |
| Health-Ade | Claude | 17 | Perplexity | 1 |
| BRĒZ | Perplexity | 24 | Claude | 0 |
| Sunwink | Claude | 26 | Gemini | 0 |
| Aplós | Gemini | 17 | ChatGPT | 0 |
| Three Spirit | Perplexity | 15 | ChatGPT | 1 |
| REBBL | ChatGPT | 19 | Gemini | 0 |
| TRIP | Gemini | 15 | Claude | 0 |
| Clevr Blends | Claude | 18 | ChatGPT | 1 |
| Seedlip | Claude | 18 | Gemini | 0 |
A platform gap is included here when a top-25 brand has at least 10 appearances on its strongest platform and no more than one on its weakest.
Each buyer situation contains 120 usable responses. The numbers beside each brand show its response appearances within that cluster.
Calm & UnwindEnergy & FocusTaste & Social OccasionIngredients & Product FitRetail & Trial
Buyer situation 1
Relaxation without alcohol, excessive stimulation or unwanted drowsiness.
Recess and Kin hold the clearest cross-platform association with calm. Moment is a meaningful third option, while the rest of the field drops sharply.
Buyer situation 2
Sustained attention, lower-jitter energy and alternatives to another coffee.
This is the least settled buyer situation. The leading brand appears in only 14 responses, and 54 responses return no reviewed packaged beverage brand at all.
Buyer situation 3
Adult flavor, food pairing, entertaining and alcohol-free social rituals.
This is where functional drinks and mindful-drinking brands collide. Ghia ranks fourth despite positioning itself as a non-alcoholic apéritif rather than a functional beverage.
Buyer situation 4
Caffeine sensitivity, format, sugar, ingredient transparency and specific functional needs.
Clear product architecture matters here. Recess, Liquid I.V., Four Sigmatic and Nuun are repeatedly matched to concrete ingredients and formats.
Buyer situation 5
Where to buy, delivery, price, variety packs and low-commitment trial.
Large retail footprints dominate. OLIPOP, Liquid I.V., Celsius and Poppi all outperform brands that lead the calmer or more social occasions.
Recess surfaced for 41 of 50 buyer questions and appeared in all 3 runs for 41 question-platform combinations. Its lead is not built only on scattered one-off recommendations. It holds across calm, social, ingredient and retail situations.
Hiyo surfaced for 33 buyer questions - second only to Recess - but placed fifth by response appearances. It appeared in all three runs for only 12 question-platform combinations and reached just 7 of 150 ChatGPT responses. Hiyo fits many situations, but the recommendation is not consistently reproduced.
Ghia appeared in 51 responses, ranked 12th overall and surfaced on 4 platforms. In Taste & Social Occasion, it ranked fourth with 36 of 120 response appearances. Ghia describes its product as a non-alcoholic apéritif. The result shows AI treating botanical complexity, low sugar and alcohol-free ritual as functional-category signals even when the brand does not claim that identity.
Claude returned an average of 6.68 reviewed beverage brands per response, compared with 2.79 for ChatGPT. ChatGPT produced 56 responses with no reviewed packaged brand, while Claude produced 11. A low ChatGPT count can reflect both a brand-specific gap and a platform-level tendency to answer with fewer brands.
The Calm & Unwind questions deliberately avoided requests about anxiety, cortisol and proven health outcomes. The answers reintroduced that language anyway.
This audit measures recommendation patterns, not causality. Three recurring forms of clarity distinguish the brands that lead different buyer situations.
Signal 01
Recess and Kin are repeatedly connected to recognizable moments: unwinding, socializing without alcohol and choosing a calmer alternative. That gives AI more than an ingredient list to match.
Signal 02
Ingredient, format and caffeine questions favor brands whose products can be distinguished precisely. General wellness language is less useful than clear differences between SKUs and use cases.
Signal 03
Retail & Trial is dominated by brands with broad, legible availability. A recommendation is easier to make when the system can connect the product to retailers, formats, price and delivery.
Start with the buying situations the business actually needs to own. A single overall visibility score hides whether a brand is understood for calm, focus, social use, ingredient fit or retail trial. The right corrective work depends on which of those associations is missing.
Make product facts easy to verify. Separate each SKU's ingredients, format, flavor, caffeine, intended occasion and availability. Keep physiological claims controlled and specific, because AI systems may intensify wellness language beyond what the brand or consumer said.
Build evidence, not volume. The 2024 KDD paper GEO: Generative Engine Optimization found that citations, relevant quotations and statistics could improve source visibility in generative responses, while the effectiveness of tactics varied by domain. For beverage brands, that argues for precise product documentation and credible third-party support rather than mass-producing generic wellness content.
Aggarwal, P. et al. (2024). “GEO: Generative Engine Optimization.” Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 5-16. DOI: 10.1145/3637528.3671900.
This audit captures what four AI platforms recommended during the August 14, 2026 test window. Results can change as models, retrieval systems and public information change.
Response appearances measure inclusion, not recommendation quality, factual accuracy, sentiment, sales impact or causal influence on purchase. Buyer-question coverage is used only as a secondary breadth measure.
Category crossovers are intentionally visible in the analysis. Their inclusion documents how AI systems define the competitive field; it does not reclassify those brands' own market positioning.