Stop taking notes in meetings

The problem

Someone in every meeting is typing instead of thinking. They participate at half capacity, produce notes that capture what they had time to catch, and send them a day later if at all. Everyone else leaves with a slightly different memory of what was decided, and the difference surfaces two weeks later as "I thought we agreed the other thing".

The action items are worse. They are spoken, everyone nods, and then they exist only in the intentions of whoever felt responsible. Nothing is written where the whole group can see it, so follow-through depends on individual conscientiousness rather than on a record.

And the archive problem compounds. A decision from three months ago is in nobody's notes, so it gets rediscussed from scratch, which is the most expensive kind of meeting there is.

What AI can actually do here

This is one of the places where the AI is genuinely doing the work rather than decorating it.

A meeting assistant joins the call, transcribes it live, identifies speakers, and produces a summary with action items afterwards. That means nobody types, everyone participates, and the record is complete rather than selective. Because transcripts are searchable, the archive becomes useful: a half-remembered decision is findable by typing what you remember of the words.

A different tool for a related problem: async video replaces the meeting that only existed to transmit information, and with AI features it summarizes itself, extracts tasks and can be edited by cutting text from the transcript.

Two honest limits. Accuracy drops with crosstalk, accents and poor audio, and AI summaries occasionally state something confidently that nobody quite said, so a summary is a draft to check rather than a record to forward blind. And recording people carries obligations: in the EU you need a lawful basis and, in practice, to tell participants clearly rather than quietly enabling a bot. Some conversations should not be recorded at all, and knowing which is part of using these tools well.

How the tools differ

Fireflies.ai meters storage rather than transcription. Its free plan gives unlimited transcription and unlimited AI summaries with 400 minutes of storage per team, then Pro at $10 per seat per month billed annually with 8,000 minutes of storage per seat. It connects to Zoom, Google Meet and Microsoft Teams, and AskFred answers questions across your meetings.

Otter.ai meters minutes. Basic is free with 300 monthly transcription minutes, Pro is $8.33 per user per month billed annually with 1,200 minutes, and Business is $19.99 with unlimited. Its live notes make it strong when you want to mark moments during the call.

Loom removes meetings rather than documenting them: a recorded walkthrough instead of a call, free for 25 videos with 5-minute recordings, and $18 or $24 per user per month for unlimited length and the AI layer. Otter.ai vs Fireflies.ai compares the two note-takers directly.

When you don't need a tool

If your meetings are two people and the outcome is one decision, write the decision in a message afterwards. That takes thirty seconds and produces a shorter, more useful record than a transcript.

If the real problem is too many meetings, a note-taker documents the waste rather than removing it. Cancelling the meeting that exists only to share updates is a bigger win, and Loom or a written update covers it.

And for sensitive conversations, do not record. Performance discussions, disagreements and anything involving personal circumstances change character when a bot is present, and the legal exposure is highest exactly there.