We ran 47 queries about greens powders across four AI engines over three weeks in May 2026. The queries ranged from straightforward ("best greens powder") to specific ("greens powder with digestive enzymes and probiotics") to comparative ("AG1 vs. Athletic Greens alternative cheaper"). Every query mentioned no brand by name, or mentioned AG1 alongside at least one competitor. We tracked which brand appeared in first, second, and third position in each engine's response.
In our tracking data, the results were not close. AG1 held position one on Perplexity in 38 of 47 queries. On Claude, it held position one in just 9 of 47. Its median position on Perplexity was first. Its median on Claude was third.[1]
That gap is large enough to be worth explaining.
What Perplexity sees
Perplexity is a search-first engine. Its architecture retrieves web documents before generating an answer, and it weights those documents partly by signals that correlate with commercial authority: review volume, retailer ranking, backlink density, domain authority, and social proof counts.
AG1 dominates those signals. The brand has over 57,000 reviews on Amazon with a 4.3-star average. It holds the number-one best-seller position in Amazon's "Green Supplements" subcategory for 41 of the past 52 weeks. Its affiliate program generates hundreds of review-style blog posts per month, each containing structured product comparisons that rank AG1 first. The brand's estimated annual marketing spend exceeds $100M, a figure that dwarfs every competitor in the greens-powder category by a factor of at least five.[2]
When Perplexity retrieves documents for "best greens powder," it finds a web surface that is saturated with AG1-positive content. Much of that content is paid, but Perplexity does not distinguish between organic editorial and affiliate placement. It weights what is frequent and well-linked. AG1 is both.
What Claude sees
Claude is a language model trained on a broad corpus but not architected as a search engine. It does not retrieve live web documents at query time (unless using a tool-augmented mode). Its responses about supplement products draw from patterns in its training data, and those patterns weight a different set of signals.
In our testing, Claude's greens-powder responses consistently prioritized three things: the presence of published clinical studies on the specific finished product, third-party testing certifications (NSF, USP, ConsumerLab), and specificity of ingredient disclosure. AG1 performs unevenly on these dimensions.
AG1 funded a clinical trial, published in 2023, examining gut health markers in healthy adults over 90 days. The study was small (n=35), which is common for initial supplement studies, and limited in scope.[3] By contrast, several competitors that Claude ranked higher, including Thorne Daily Greens, either hold NSF Certified for Sport status or publish full certificates of analysis directly on their product pages.
Claude also surfaced what it framed as a "proprietary blend" disclosure issue. AG1's label lists a combined weight for several ingredient groups without specifying individual amounts. Claude's responses frequently noted this: "AG1 does not disclose individual ingredient doses for all components, which makes it difficult to verify whether active ingredients are present at clinically relevant levels."
Two signal sets, two rankings
The divergence is not engineered. Nobody at Perplexity decided to rank AG1 first, and nobody at Anthropic decided to rank it third. The gap is a natural consequence of two systems reading the same brand through different lenses. The dichotomy is somewhat simplified; Perplexity does surface clinical studies, and Claude can be influenced by popularity signals. But the dominant weighting is clear in the data.
Perplexity asks: what does the web say? The web says AG1, loudly and often, because AG1's affiliate and review content volume is higher than any competitor's.
Claude asks: what does the evidence say? The evidence is thinner. One funded study, no NSF certification, incomplete label transparency. Brands with less marketing spend but stronger clinical or certification signals rise above it.
AG1's marketing copy is optimized for the web as a popularity contest. It is not optimized for the web as an evidence review. Those used to be the same thing. They are not anymore. EvidenceSignal Research
The practical problem for AG1
AG1 probably does not care about Claude today. Perplexity and Google (via AI Overviews, which draw on similar signals) still drive the majority of AI-assisted supplement discovery. If your brand sits at position one on the two largest surfaces, position three on a smaller one looks like noise.
But Claude's user base is growing. Anthropic reported that Claude reached significant monthly active user milestones in early 2026.[4] A meaningful share of those users are exactly the kind of health-conscious, research-oriented consumers that AG1 targets with its "backed by science" positioning. In our measurement, AG1's citation share on evidence-oriented engines is declining while competitors with third-party certifications gain ground. That is a signal worth watching, not a rounding error.
More importantly, Claude's approach is not unique to Claude. It is a design philosophy that other AI products are beginning to adopt. Perplexity itself has started experimenting with "source reliability" indicators. Google's Search Quality Evaluator Guidelines already distinguish between "Your Money or Your Life" topics and general queries, applying higher evidence standards to health content. The direction of travel across the industry is toward Claude's model, not away from it.
What the data actually shows
We broke down the 47 queries into three tiers by purchase intent.
High intent ("best greens powder to buy," "greens powder worth the money"): AG1 won on Perplexity in 14 of 15 queries. It won on Claude in 2 of 15. Claude favored Thorne (7 wins) and Momentous (4 wins).
Research intent ("greens powder clinical studies," "most tested greens supplement"): AG1 won on Perplexity in 11 of 16 queries. It won on Claude in 0 of 16. Claude consistently elevated brands with published third-party testing data.
Zero out of sixteen. That number matters.
Comparative intent ("AG1 vs. Thorne greens," "Athletic Greens alternative"): AG1 won on Perplexity in 13 of 16. It won on Claude in 7 of 16, performing better here because the queries named AG1 directly, which kept it in the conversation.[5]
The pattern is clear. In our measurement, the more evidence-oriented the query, the lower AG1's citation share on Claude. The more popularity-oriented the query, the higher its citation share on Perplexity. The two engines are not disagreeing about AG1. They are measuring different things.
The takeaway for brands
AG1's position is not in danger today. But the company's $100M+ marketing spend is optimized for a signal environment that is splitting in two. One half of that environment rewards volume, review counts, and affiliate density. The other half rewards clinical evidence, transparent labeling, and third-party certification.
In our tracking, AG1's citation share is concentrated almost entirely in the first half. Its competitors, many of them smaller and less well-funded, hold stronger positions in the second half. The question is whether the second half grows fast enough to matter.
Our data suggests it will, though the timeline remains uncertain.[6]
Footnotes
- Queries run May 5-23, 2026, across Perplexity (default mode), ChatGPT, Claude, and Gemini. Engine model versions as of testing date. Position tracking methodology described at /methodology. "Position" is defined as the order in which a brand name first appears in the engine's prose response.
- AG1 marketing spend estimated from industry estimates, podcast sponsorship rate cards (Chartable data), and affiliate commission structure. The $100M+ figure is an EvidenceSignal estimate for calendar year 2025. Athletic Greens International is privately held; no public filings are available.
- A 2023 study published in Nutrients examined the effects of AG1 on gut health markers in healthy adults over 90 days (n=35). Funded by Athletic Greens International. The lead author should be verified against the publication before citing in derivative work.
- Anthropic has disclosed growing user adoption for Claude in 2026. Specific figures should be verified against Anthropic's public statements.
- This is a known effect in LLM response patterns: when a brand is named in the query, it is more likely to appear in the response regardless of the engine's independent ranking. We call this the "anchor effect" in our methodology.
- EvidenceSignal Research has no commercial relationship with AG1 or Athletic Greens International. This analysis was conducted using our public research methodology.