Which AI note-takers do ChatGPT and Perplexity recommend?
Twelve buying questions, asked of both engines on the same day. They did not agree.
- Answers read
- 24
- Tools named
- 56
Buyers no longer open ten tabs to pick a meeting-notes tool. They ask an assistant and work from the shortlist it gives them. That shortlist is not a search ranking, and nobody publishes it, so we measured it: twelve questions a real buyer would type, asked of ChatGPT and Perplexity, with every product named in every answer counted.
Named most often
Fireflies.ai
12.2% share of voice, named in 15 of the 24 answers.
Listed first most often
Fathom
Opened 4 answers against Fireflies.ai's 1. Fewer mentions, better placement.
| # | Tool | Share of voice | ChatGPT | Perplexity | Avg. position |
|---|---|---|---|---|---|
| 1 | Fireflies.ai | 12.2% | 7 | 8 | 3.1 |
| 2 | Fathom | 8.9% | 5 | 6 | 2.2 |
| 3 | Otter.ai | 8.1% | 4 | 6 | 3.2 |
| 4 | Fellow | 5.7% | 4 | 3 | 2.1 |
| 5 | Granola | 5.7% | 2 | 5 | 3.9 |
| 6 | Jamie | 4.1% | 1 | 4 | 2.6 |
| 7 | tl;dv | 4.1% | 0 | 5 | 3.8 |
| 8 | Notion | 2.4% | 1 | 2 | 1.7 |
| 9 | Shadow | 2.4% | 0 | 3 | 2.3 |
| 10 | Zoom AI Companion | 2.4% | 1 | 2 | 3 |
| 11 | Krisp | 2.4% | 2 | 1 | 4.3 |
| 12 | Dovetail | 1.6% | 1 | 1 | 1.5 |
| 13 | Circleback | 1.6% | 0 | 2 | 3 |
| 14 | Notta | 1.6% | 0 | 2 | 3.5 |
| 15 | Sembly AI | 1.6% | 0 | 2 | 4.5 |
| 16 | Descript | 1.6% | 1 | 1 | 5 |
| 17 | HappyScribe | 1.6% | 0 | 2 | 5 |
A further 39 tools were named exactly once across the run. One mention is inside the margin, so they sit below the table rather than in it.
The two engines are not reading the same internet
ChatGPT
Opened 7 of 12 answers with a tool nothing else in the run named.
Perplexity
Opened 1 of 12 answers with a tool nothing else in the run named.
Both engines searched the live web for every one of these questions, on the same day, with the same wording. They still came back with different categories of answer. ChatGPT opened with small, heavily optimised products and cited them directly: hynote.ai, minutewhisper.com, joinbaxter.com, askiyo.com. Perplexity opened with tools that are in the table above.
6 ranked tools were named by Perplexity and by no other engine: tl;dv, Shadow, Circleback, Notta, Sembly AI, HappyScribe.
For a brand, that is the practical takeaway on this page. Being the known name in a category is what earns Perplexity’s shortlist. It is not what earns ChatGPT’s, which went to whoever had a fresh, crawlable page answering the question directly.
What the engines were reading
Sources cited while answering these questions, by how many answers leaned on them. This is the shortest route into the category: get covered here and you get named there.
Question by question
The table is the category. This is the part that matters to a single brand: which tool each engine reached for first, on each question. Picks flagged “named only here” appear in that one answer and nowhere else in the run.
What's the best AI note taker for meetings?
Best AI meeting assistant for a small team
Which AI tool should I use to transcribe and summarise Zoom calls?
Recommend an AI notetaker that works with Google Meet
What's the best alternative to Otter.ai?
Best AI meeting notes tool for sales teams
Which AI meeting assistant is best for a privacy-conscious team?
Cheapest AI meeting transcription tool for recurring calls
Best AI notetaker that integrates with Notion
What AI tool automatically writes meeting summaries and action items?
Best AI meeting assistant for consultants and agencies
Best AI notetaker for user research interviews
How we measured this
On 29 July 2026 we asked ChatGPT and Perplexity each of the 12 questions above, once, with live web search on. That produced 24 answers. Every product an answer put forward was recorded in the order it appeared, so a tool that opens an answer counts differently from one mentioned at the end.
Share of voice is that tool’s slots as a percentage of every slot in the run. Answer engines and the sites the answers cited are excluded from the tally: Reddit is where an answer came from, not a meeting-notes tool. Spellings of the same brand are merged, and the spelling the engines used most is the one shown.
The limits are worth stating. One run is one observation, and the same question next month can return a different shortlist. Twelve questions cover the common buying paths, not every niche. And a tool named once is inside the margin, which is why those sit below the table rather than in it.
Questions, answered
A listicle is one writer's opinion. This is a count. We ask ChatGPT and Perplexity the same twelve buying questions, record every product each answer names and in what order, and publish the tally next to the questions themselves. You can argue with the questions, since they are printed on this page. You cannot argue with the arithmetic.
Because it searched, and those pages answered the question. ChatGPT ran a live web search for every question here and repeatedly opened with small products whose sites it cited directly. Those are real companies with real sites, not invented ones. They win the slot by having a fresh, crawlable page aimed squarely at the question, which is a different qualification from being the known name in the category. Perplexity, given the same questions on the same day, mostly named the established tools.
Every answer names a handful of tools. Add up all of those slots across all answers, and share of voice is the percentage a given tool occupies. A tool at 12% appeared in roughly one in eight of the recommendations the engines made in this category.
Those are the two answer engines we measure live for customers, so they are the two we can measure honestly here. Google's AI Overviews are not included. We would rather rank two engines with a real method than four with a guess.
Yes, and that is the point. AI answers are not a fixed index. The same question asked next month can return a different shortlist. This page shows one dated run. We re-run it and update the table rather than quietly leaving old numbers up.
Then the engines did not name it for these questions, which is a finding rather than an oversight. Run your own brand through the free check to see which questions you show up in, and which sources the engines are reading instead.
This is the same measurement, run on your brand.
Your questions instead of ours, your category instead of this one. Free, about thirty seconds, and it names whoever the engines picked over you.