We Asked Five AI Systems About Meeting Assistants. The Answer Changed With the Question.

Fathom and Fireflies.ai appeared equally often across the 59 successful responses to questions that did not name a brand: 35 mentions each.

Otter.ai appeared in 24 responses, Avoma in 21, and tl;dv in 15.

These are mentions, not endorsements. One answer can include several brands, and a brand can appear as a comparison or alternative rather than the preferred choice.

The figures are a snapshot of this particular prompt set, not market share or a universal visibility ranking.

The most consistent result concerned a growing team seeking searchable meeting history, workflow integrations and a free starting plan.

All five AI systems placed Fireflies.ai first or presented it as their leading choice for that scenario.

However, when buyers asked about sales coaching and CRM updates, the answers split: Gong led two responses, Avoma two, and Fathom one.

This is a qualitative reading of one response per engine and question. A first-listed item does not always represent an explicit ranked recommendation.

The table below shows the first prominently recommended or first-listed brand in each observed response, based on qualitative review. These are not objective product winners.

1. Category visibility and buyer-intent visibility are different.

A brand might appear often in broad answers but lose ground when the user adds a specific requirement, such as CRM workflows, team-wide search or sales coaching.

Those more specific questions can be closer to a real buying decision.

2. A single score can hide the reason a brand wins or loses.

Teams should investigate the questions, use cases and AI models driving an apparent visibility gap.

The most valuable question is not only “Did we appear?” but “Did we appear when a buyer described the problem our product is meant to solve?”

3. Count mentions carefully.

We reviewed the underlying response text rather than treating automatic classification as ground truth.

Some generated answers included image captions, URLs and competitor tags that could distort a simple count.

The mention figures presented here were derived by checking brand names in the response text after removing image markup and raw URLs. The example buyer-scenario outcomes were checked against the underlying answers.

This exploratory analysis used an existing Conviora Full Report for fathom.video configured for the United States and English on November 4, 2026.

The Full Report generated 15 questions across five AI systems.

Three questions explicitly named Fathom and were excluded from the cross-brand analysis.

The remaining 12 questions produced 60 expected responses. Of these, 59 succeeded and one ChatGPT observation failed.

All observations came from the same measurement window.

The AI systems were ChatGPT, Claude, Gemini, Google AI Mode and Perplexity.

The recorded models, where available, were GPT-6, Claude Sonnet 5, Gemini 3.5 Flash-Lite and Perplexity Sonar. The Google AI Mode model identifier was not recorded.

The measured provider configurations may differ from the corresponding consumer applications.

This was one measurement run, not a repeated or statistically representative study.

The brand-neutral questions were generated as part of a Fathom-focused report, so topic selection may favor particular brands or use cases.

Mentions do not prove product superiority, user preference or actual market adoption.

AI responses change over time.

A controlled multi-brand benchmark would require a fixed, independently defined question set, repeated measurements and validated answer labeling.

Explore the underlying Fathom Full Report

Published by Conviora Research. This is independent exploratory research, not a paid endorsement or certification of any company mentioned.