Last week, Recognized, Not Recommended described a pattern: two AI assistants, the same brand, the same category, the same day, and very different answers. The brands were not named. Readers asked two fair questions: how common is this, and who are the brands?
This piece answers both. It is not a new claim. It is the data behind the earlier one.
Most teams come to AI visibility assuming there is one shelf: one place where buyers using AI either find you or don't. The data says otherwise. There are four shelves, one per AI assistant, and each is stocked differently. Not slightly differently. Differently enough that your place on one tells you little about your place on the others. The earlier examples were not edge cases. They are how the shelves work.
What follows is one week (the week of June 28, 2026) from the Envoyra observatory. We asked ChatGPT Search, Claude, Gemini and Perplexity the same buying questions across several software categories. For each brand, we counted how often it was named on the questions in its own category.
The brand on every shelf, unevenly
Start with the least dramatic case, because it sets the baseline.
HubSpot, in small-business CRM, was named in most answers on all four AI assistants. It was named most on ChatGPT Search and least on Perplexity, but the spread was narrow. Zendesk, in customer support, looked much the same: named in most answers everywhere. So did ActiveCampaign in marketing automation.
That is what four-shelf coverage looks like: present everywhere, within a narrow band. It is the position every brand assumes it has. That week, few brands actually had it.
The brand on one shelf and half of another
Pipedrive was named in every Claude answer to small-business CRM buying questions that week. On Gemini, it was left out of many.
Same brand. Same week. Same questions. One AI assistant treated Pipedrive as part of every answer. Another often left it out.
Klaviyo, in marketing automation, was named in most answers on Claude and far fewer on ChatGPT Search. A buyer asking Claude meets Klaviyo as a fixture of the category. A buyer asking ChatGPT Search meets it now and then.
Salesforce, in customer support, was named often on Claude and rarely on ChatGPT Search. Fireflies, in meeting assistants, was named in most Claude answers but far fewer on ChatGPT Search and Gemini.
One AI assistant treated Pipedrive as part of every answer. Another often left it out. Same brand, same week, same questions.
The single-shelf brands
Then there are brands that, that week, lived on just one shelf.
Brevo was named often on Perplexity for marketing automation questions, and almost never on ChatGPT Search. Granola, a newer meeting assistant, was named on Perplexity and not at all on ChatGPT Search or Gemini. Read AI was named on Perplexity and rarely, if ever, anywhere else.
For a buyer who uses Perplexity, these brands are part of the category. For a buyer who uses ChatGPT, they do not exist. Neither buyer knows the other's shelf is different.
Why one combined number hides this
Average Klaviyo's four results and you get a comfortable middle figure that matches none of its four realities. Averaging across AI assistants erases the finding. Checking just one does the same. A team that tested Pipedrive on Claude that week would have concluded it was named everywhere. A team that tested on Gemini would have raised the alarm.
Both teams would have been describing one shelf and believing they had described the whole store.
We are not explaining these gaps in this piece. Each AI assistant builds answers from its own sources, on its own schedule. We did not measure why any one brand leans one way. What we did measure is the shape, and the shape repeats.
Across all four June weeks, the same brands leaned the same way. Klaviyo was named far more on Claude than on ChatGPT Search every week. Brevo was named far more on Perplexity than on ChatGPT Search every week. Granola was never named on ChatGPT Search in any of the four weeks. The gaps are steady enough to be part of each brand's position, not noise in the measurement.
The question to take away
"Are we visible in AI?" has no answer, because it assumes one shelf.
These questions do have answers. Which AI assistants name us, how often, and on which buying questions? Which AI assistants do our buyers actually use? And is the gap between our best shelf and our worst shelf growing or shrinking?
A brand that can answer those three knows where it stands in AI. A brand with a single number has an average.
**Method notes.** Findings come from the Envoyra observatory run of June 28, 2026. We asked four AI assistants the same buying questions across several software categories, and received almost every expected answer. For each brand, we counted the answers that named it, out of the questions in its own category. Claims about week-to-week consistency compare the four weekly June runs (June 7, 14, 21 and 28) on the AI assistants with complete data in each week. This is one week's snapshot. Positions move over time, and we track that movement separately in this series. One category we measure (AI visibility tools) is left out of the examples because Envoyra competes in it. The underlying answer data is kept and can be audited.
*This article is part of The Invisible Shelf™, a series by Envoyra co-founder and CEO Golara Serio on AI recommendation systems, brand discovery, and the measurement infrastructure being built in real time.*