AI meeting notetakers vs writing it yourself - what you lose and gain

In the last two years, AI notetakers have become normal. Teams, Zoom and Google Meet offer built-in AI summaries on certain plans, and third-party bots such as Otter, Fireflies and others join calls, record, transcribe and send a summary afterwards. Many people now finish a meeting with a tidy list of action items they did not write.

So is typing your own notes obsolete? Not quite. Here is what I think the trade actually is.

What AI notetakers do well

• Completeness. A transcript does not miss a sentence because you were thinking.
• Searchable record. “What did we say about the budget in March?” becomes answerable.
• Summaries for people who missed it.
• No typing for people who find note-taking distracting.

For long, information-heavy meetings, a recording and summary is genuinely useful.

What they miss

Your private follow-ups. A summary lists what was said. It does not know what you privately decided to do because of it: “check whether this affects the Leeds contract”, “ask Sam about this one-to-one”.

Judgement. Summaries flatten emphasis. The offhand comment that will matter most often appears as one bullet among twenty.

Ownership. An action item in a summary is not yet a task in your system. It still needs moving somewhere with a date.

Accuracy on names and numbers. Transcription has improved enormously, but names, figures and jargon still go wrong. A summary with a wrong number is worse than no summary.

The consent problem

Recording a conversation has legal and social implications. Many jurisdictions require consent from participants, and workplaces increasingly have policies on AI bots joining calls; some organisations block them entirely. External participants may not expect a bot to be listening. The UK ICO’s guidance on AI and data protection is a reasonable starting point for the data side.

Typed notes have no such issue. They are your notes.

What I actually do

A hybrid:

1. Let the recording exist when the meeting needs one and everyone has agreed.
2. Type my own lines anyway, only for things that are mine: decisions that affect me, actions I owe, questions I want answered.
3. After the meeting, check my lines against the summary, fix any wrong numbers, and turn actions into dated tasks.

The typed lines take maybe ten seconds per meeting. They are the part that turns a meeting into something I do.

Where Jot fits

Jot’s Meeting tab, coming in the next release, is built for the second step. You type a line per point, and Jot sorts lines into decisions, actions for named people, notes and open questions. Your own actions become todos with dates. It records nothing and sends nothing to a transcription service. See minutes nobody has to tidy and parsing a meeting line.

It also hides itself from screen sharing by default, so typing during a presentation is not broadcast. See hidden from the screen you share.

The short version

AI notetakers are great at remembering what was said. You are still the only one who knows what it means for you. Write that part down.

Jot is on the Microsoft Store.

Jot is a quick-capture app for Windows: one hotkey, one line, back to what you were doing. What it is, or what is still unproven.