Last updated: 10 August 2026
Ask ChatGPT, Google AI Overviews or Gemini which platform to use for enterprise AI search, AI agent building or workflow automation, and Notion AI comes up most often, named in 17.9% of the 402 answers we analysed and taking 32.8% of all brand mentions in them. Glean is close behind at 17.2%. But no single brand owns this category: the leader accounts for about a third of the mentions, and the rest is split across 14 other names. The finding least likely to be visible from the inside is not who wins, but who is spending money to be seen here. Across the 55 questions we track, 24 distinct brands buy ChatGPT ads, and not one of them is a brand that leads the organic answers.
Methodology: what we tracked and how we counted it
We track 55 questions that buyers plausibly ask an AI engine about enterprise search, AI agent building and workflow automation. Each one runs against ChatGPT, Google AI Overviews and Gemini, every day. The leaderboard, sentiment, position and source figures in this article are drawn from 402 answers collected between 8 and 10 August 2026.
We record what the engine actually returns rather than what it is asked. For every answer we log which brands are named, where in the answer each one appears, how the answer characterises them, and every source URL the engine retrieved to build it. No brand is credited for appearing in a question. Only what comes back in the answer is counted.
Two metrics carry the leaderboard. "How often named" is the share of the 402 answers that mention the brand at all. "Share of the answer" is that brand's slice of all brand mentions counted across the whole set, so how much of the category conversation belongs to it. It is an aggregate across 402 answers, not a description of any single one. Sentiment is scored 0 to 100. Average position records where in the answer the brand tends to appear when it is named; lower means earlier.
The leaderboard: 15 brands, no clear owner
| Brand | Named in % of answers | Share of the answer | Sentiment /100 | Avg. position when named |
|---|---|---|---|---|
| Notion AI | 17.9% | 32.8% | 55 | 2.3 |
| Glean | 17.2% | 20.4% | 56 | 2.3 |
| Microsoft 365 Copilot | 15.4% | 9.6% | 59 | 3.5 |
| Zapier | 10.7% | 8.7% | 59 | 4.1 |
| LangChain (incl. LangGraph) | 9.2% | 7.5% | 58 | 3.8 |
| n8n | 7.2% | 4.5% | 59 | 5.3 |
| Guru | 5.5% | 4.3% | 57 | 3.4 |
| MindStudio | 5.5% | 3.1% | 61 | 3.8 |
| GoSearch | 3.7% | 2.5% | 60 | 3.3 |
| Relevance AI | 3.2% | 1.5% | 66 | 3.5 |
| Lindy | 2.7% | 2.5% | 61 | 2.2 |
| Gumloop | 2.5% | 1.2% | 64 | 6.7 |
| Dust | 1.7% | 1.0% | 65 | 6.3 |
| StackAI | 1.0% | 0.5% | 61 | 5.0 |
| Moveworks | 0.25% | 0.2% | 53 | 7.0 |
Three things stand out once the table is in front of you.
Nobody owns the category. Notion AI leads on both frequency and share, but its 32.8% means that across the whole set, roughly two thirds of all brand mentions belong to someone else. Add Glean's 20.4% and the top two together account for 53.2%, leaving 46.8% split across the other 13 brands. This is a fragmented category, not a two-horse race.
The most-named brands do not draw the warmest sentiment. Notion AI and Glean, named more often than anyone, sit near the bottom of the sentiment table at 55 and 56. Moveworks posts the single lowest score, 53, but it was named in only 0.25% of the 402 answers, so that figure rests on very little evidence. The clearer pattern sits at the top: the three highest sentiment scores, Relevance AI (66), Dust (65) and Gumloop (64), all belong to brands named in under 4% of answers. Broad category coverage does not buy warm coverage. A narrow, well-defined tool draws cleaner praise than a brand an engine has to describe across many different use cases.
Being named is not the same as being recommended. Position tells a different story to frequency. Lindy is named in just 2.7% of answers, less than a sixth of Notion AI's rate, yet when it does appear it sits earlier than any other brand in the table, at position 2.2 against Notion AI's 2.3. Gumloop and Dust are named at broadly similar rates, 2.5% and 1.7%, but appear far later, at positions 6.7 and 6.3. The same split shows up higher in the table: n8n is named in 7.2% of answers at position 5.3, while Guru is named less often at 5.5% but appears at 3.4. Frequency and position are measuring different things, and only one of them tells you where in the answer a buyer meets the name.
Who is buying ChatGPT ads in this category
Across the 55 questions we track, 24 distinct brands ran ChatGPT ads. Ads appeared on 36 of the 55 questions, an overall ad coverage rate of 10% of tracked responses, and each advertiser targeted an average of 2.4 questions.
Share of ad exposure
| Advertiser | Share of ad exposure |
|---|---|
| SnapLogic | 18% |
| Tidio | 13% |
| Trackerly | 12% |
| Freshworks | 10% |
| UiPath | 6% |
| Monday.com | 5% |
| Board Intelligence | 4% |
Advertiser detail
| Advertiser | Creatives | Topics | Spend tier | Times seen |
|---|---|---|---|---|
| SnapLogic | 2 | 3 | High | 15 |
| Tidio | 1 | 2 | High | 11 |
| Trackerly | 1 | 1 | High | 10 |
| Freshworks | 2 | 4 | High | 8 |
| UiPath | 2 | 2 | High | 5 |
| Monday.com | 1 | 2 | High | 4 |
| Board Intelligence | 1 | 1 | High | 3 |
| CloudTalk | 2 | 1 | High | 3 |
| Mitratech | 2 | 2 | Medium | 3 |
| NiCE | 2 | 2 | Medium | 3 |
| Layer AI Systems | 1 | 2 | Medium | 3 |
| Datasite | 1 | 1 | Medium | 2 |
| Jotform | 2 | 2 | Medium | 2 |
| AlphaSense | 1 | 1 | Medium | 1 |
| Client Ascension | 1 | 1 | Medium | 1 |
| Cursor | 1 | 1 | Medium | 1 |
| 8 further advertisers (E-Sign, Ficode Technologies, Greymill AI, Keeper Security, Laird Assessors, Toloka AI, Twilio EMEA, ManageEngine) | not recorded | not recorded | Low | 1 each |
Now put this table next to the leaderboard above. None of the nine brands that actually lead the organic answers, Notion AI, Glean, Microsoft 365 Copilot, Zapier, LangChain, n8n, Guru, MindStudio or GoSearch, bought a single ad. Every one of the 24 advertisers sells into an adjacent problem rather than competing for the category itself. SnapLogic, the largest advertiser by exposure share, does not appear anywhere in the 15-brand leaderboard. Neither does Tidio, Trackerly, UiPath, Monday.com or Board Intelligence.
Ad coverage is not even across the category's sub-topics. Questions about enterprise search and knowledge carried ads 17% of the time, AI agent building 15%, and workflow automation 11%.
Two creatives, recorded verbatim, give a sense of how advertisers are positioning against this attention. Neither company appears anywhere in the leaderboard above:
Greymill AI: "The AI for Small Businesses. One AI team, industry specific knowledge. 16 Specialists. Free 7-day trial."
Tidio, the second largest advertiser by exposure: "Custom Enterprise AI Agents. From AI audit to full implementation."
Where these answers actually come from
AI engines do not invent these recommendations. They retrieve sources and build the answer from what they read. Across the 402 answers, we counted every domain the engines pulled from and how often each one appeared.
| Rank | Domain | Retrievals | % of answers it appeared in |
|---|---|---|---|
| 1 | microsoft.com | 75 | 18.7% |
| 2 | youtube.com | 69 | 17.2% |
| 3 | reddit.com | 48 | 11.9% |
| 4 | medium.com | 45 | 11.2% |
| 5 | linkedin.com | 44 | 10.9% |
| 6 | arxiv.org | 40 | 10.0% |
| 7 | glean.com | 39 | 9.7% |
| 8 | gosearch.ai | 34 | 8.5% |
| 9 | openai.com | 33 | 8.2% |
| 10 | slack.com | 30 | 7.5% |
| 11 | mindstudio.ai | 28 | 7.0% |
| 12 | salesforce.com | 27 | 6.7% |
| 13 | google.com | 27 | 6.7% |
| 14 | ibm.com | 24 | 6.0% |
| 15 | coworker.ai | 22 | 5.5% |
| 16 | techradar.com | 22 | 5.5% |
| 17 | mckinsey.com | 21 | 5.2% |
| 18 | zapier.com | 21 | 5.2% |
| 19 | notion.com | 19 | 4.7% |
| 20 | nist.gov | 18 | 4.5% |
Domain type split across all retrievals: corporate 65%, user-generated content 9%, other 7%, institutional 7%, competitor 6%, editorial 4%, reference 3%.
Three patterns are worth pulling out. YouTube is the second most-retrieved source in the whole dataset, ahead of Reddit at third. Video and community discussion outrank most corporate blogs and vendor sites for this category, which says something about how buyers actually research AI tools: they watch someone use the product before they read a vendor's copy about it. Arxiv.org sits sixth, an unusually high placement for a category that has plenty of buyer-facing content available. That tells us these answers lean on genuinely technical material, not just marketing pages, when engines explain what an AI agent platform or enterprise search tool actually does.
The third pattern is the one with the clearest practical read. Five vendor-owned domains sit inside the top 20 sources: glean.com at seventh, gosearch.ai at eighth, mindstudio.ai at eleventh, zapier.com at eighteenth and notion.com at nineteenth. Those are not third-party write-ups. Those are the vendors' own websites, retrieved and read directly by the engines building these answers. It is a tactic working in plain sight: a well-built, technically substantive vendor domain gets pulled into the answer alongside Reddit threads and arXiv papers, not filtered out in favour of them.
What actually works in this category
Four things follow directly from the data above.
Video and community content earn their place in the source mix, not just corporate pages. Across the 402 answers, YouTube and Reddit rank second and third among all retrieved sources, ahead of most individual corporate domains. A category this technical is researched by watching and by reading practitioner discussion, so a written landing page alone is competing for a much smaller share of attention than it might assume.
A vendor's own domain is a viable citation path when the content is substantive enough. Five vendor domains already sit in the top 20 sources purely on the strength of their own content being retrieved directly. That is proof, not theory: comparison pages, technical documentation and detailed product explanations on a vendor's own site get pulled into AI answers when they carry enough specific, factual detail to be worth retrieving.
Where you land in the answer is measured separately from how often you land in it. Lindy, Gumloop and Dust are all named in under 3% of the 402 answers, yet Lindy opens at position 2.2 while the other two arrive at 6.7 and 6.3. Guru is named less often than n8n and still appears nearly two places earlier. On this data the brands that appear early are the narrower, more specifically defined tools, which is worth testing against your own positioning before assuming mention count is the number to chase.
Paid and organic are, right now, two different populations of company in this category. None of the nine brands leading the organic answers bought a ChatGPT ad across the 55 questions we track, and all 24 advertisers are absent from the leaderboard. The paid surface in this category is currently occupied by companies that do not win it organically, and the organic leaders are not contesting that surface at all.
The same analysis, run on your category
We track this same data (leaderboard, sentiment, sources and ad spend) for other categories too, and we're happy to share what a category looks like when someone runs the numbers on it.
