English is not my first language, so I used a translation tool to clean up the wording. The opinions are based on my actual experience using these tools.
I’ve been looking at a few meeting productivity tools lately because the category is starting to feel crowded. At first they all looked like some version of “AI meeting notes,” but after comparing them more closely, they seem to be solving slightly different problems.
Granola felt the most like a personal AI notepad. The landing page is very focused on taking rough notes during the meeting, then letting AI clean them up with the transcript afterward. I like that framing because it does not try to replace your thinking completely. It feels best for people who already take notes but want cleaner summaries, better recall, and less manual cleanup.
Fathom felt more execution-oriented. It is still about notes and summaries, but the positioning is more team workflow: searchable transcripts, action items, follow-ups, integrations, and keeping decisions visible across calls. I can see it fitting customer calls, sales, CS, or any team where the meeting needs to turn into next steps quickly.
Otter felt the most transcript-first. The strongest use case seems to be capturing live conversations, turning them into searchable knowledge, and letting people ask questions across past meetings. I would probably think of it first when the transcript itself matters, like interviews, lectures, internal knowledge, or situations where people need to revisit exactly what was said.
Fireflies felt broader and more enterprise-like. It covers recording, transcription, summaries, search, integrations, conversation intelligence, and even real-time suggestions/coaching. It seems useful if a team wants a central meeting archive with analytics and workflows on top. The tradeoff is that it can feel like a bigger system, not just a lightweight note tool.
Relly was the roughest one because it is still beta, but it is aiming at a different part of the meeting. Instead of only summarizing after the meeting, it tries to show draft artifacts during the meeting, like a spec, research card, or rough product direction. The idea is that people can react to the same concrete thing while the meeting is happening, instead of discovering later that everyone had a different mental picture.
My takeaway is that “AI meeting notes” is too broad as a category. Some tools are better for personal notes, some are better for follow-up work, some are better as searchable archives, and some are starting to push closer to shared outputs during the meeting itself. The right choice probably depends less on which tool has the longest feature list, and more on where your meetings usually break down.