Should I use an AI notetaker for 1:1 meetings?

Sometimes — and the answer depends much more on which 1:1 than on which tool.

Mechanically there’s nothing to discuss. A 1:1 is an ordinary video call with a link, a recurring calendar event, and audio coming out of your speakers, so a notetaker bot can join it, the platform’s own AI can summarize it, and system-audio capture works on it exactly as it works on a standup. Every method in the category handles the format.

What’s different is the format itself. A 1:1 is the only recurring meeting on most calendars that is a conversation rather than a broadcast: two people, both talking, no audience, and no way to be quietly absent from it. Three things follow from that, and they point in different directions.

• Attention isn’t the problem here. You can’t drift through a meeting where you’re half the room. The usual reason to want help during a call — you zoned out and got called on — mostly doesn’t apply, so the case for capture has to be made differently rather than borrowed.
• Note-taking costs more than it does anywhere else. In a twelve-person status call nobody notices you typing. In a 1:1 you are the entire audience for the person speaking, and looking down at a keyboard is read as withdrawal — the same effect the job interviews page describes, except this is a relationship that continues every fortnight.
• Capture can change what gets said. This is the one that matters, and it’s the reason this page exists.

The cost is paid by the other person, and you never see the invoice

In almost every meeting on this site, turning on a notetaker costs nothing to the conversation. A sprint review is the same sprint review whether or not it’s being summarized. A 1:1 isn’t, because a 1:1 earns its place on the calendar almost entirely through things people say because it’s private: I’m struggling with this. I don’t think this project is going to work. I’m not sure I want this job.

Nobody ever tells you they withheld one of those. They talk about the roadmap instead, the meeting seems fine, and the notes are excellent. The failure mode of capturing a 1:1 is a good summary of a conversation that didn’t happen.

That’s not an argument against ever capturing one. It’s an argument for knowing which conversation you’re in, because “1:1” covers several meetings that have almost nothing in common.

Type of 1:1	What it’s mostly for	Worth capturing?	The cost to watch
Cross-team / peer sync	Status, dependencies, decisions, who owns what	Yes — the easy case	Almost none; it’s a working meeting between equals
Client or customer 1:1	Requirements, commitments, follow-ups	Yes	They’re a guest on your capture; tell them, and expect a firmer standard than internally
Skip-level	Context, orientation, occasionally candid feedback	Depends	The candor is the point of a skip-level, and it’s the first thing to go
Manager ↔ report, project half	Blockers, priorities, what’s due	Usually yes	Low, if it stops before the other half
Manager ↔ report, career half	Growth, feedback, pay, doubts, personal context	Usually no	High and invisible — this is the conversation capture damages
Mentoring / coaching	Advice you’ll want to remember precisely	Yes, if the mentee agrees	Low; the person being recorded is usually the one who wants the record

The pattern is that the factual half of a 1:1 is the same as any other meeting and benefits from the same tools, while the half that makes a 1:1 worth having is the half that gets quieter when something is listening. Most 1:1s contain both, usually in that order — which is a useful thing to know, because it means the sensible default is often to capture the first part and stop.

The recurring-invite trap

There’s a structural problem specific to this meeting type, and it’s the one most likely to cause an actual incident.

1:1s are recurring events, and most bot-based notetakers work from your calendar: you connect it once, and the bot joins everything on it from then on. So the decision to capture a 1:1 gets made once, months ago, possibly by accident, and then repeats every fortnight — including the week the conversation turns into a resignation, a health disclosure, or a complaint about someone else on the team.

Nobody re-consents to a recurring meeting. The person on the other side agreed, at most, to a general arrangement in March, and has no idea it’s still running in September. If you take one practical thing from this page, make it this: a recurring 1:1 should not be on a blanket auto-join list. Turn capture on per meeting, or leave that series off entirely. How to stop AI notetakers joining your meetings covers the mechanics if one is already in the habit.

A meeting bot makes this worse in a way that’s specific to the format. Everywhere else, the argument against a bot is that it’s an extra participant, it’s blockable, and the record lives in someone else’s account. In a 1:1 it’s blunter than that: the meeting has two people in it, so a third entry named “Notetaker” isn’t a background process, it’s a chaperone. A person who wanted to raise something difficult now has to ask their manager to remove a participant first. Almost nobody will.

To be fair to bots: at least it announces itself. Bot-free capture removes the third participant and removes the announcement with it, which makes the disclosure entirely your job — and in a 1:1 that job is harder than usual, for reasons worth being precise about.

Disclosure in a two-person meeting is a negotiation, not an announcement

“Just tell people” is the standing advice on this site, and it works because in a group, “I run an AI notetaker on these” is genuinely an announcement — said to eight people, any of whom can object at low cost.

In a 1:1 there is no group. The sentence is directed at exactly one person, and if that person reports to you, they are being asked to object to their manager, alone, about their manager’s tooling, at the start of the meeting where their performance gets discussed. Asymmetric consent is the same problem the interviews page describes, and here it recurs indefinitely.

So the rule that works is to give the switch to the person with less power:

• Raise it outside the 1:1. In a message beforehand, or at the end of a previous one — not in the opening seconds, when saying no means derailing the meeting they came to have.
• Make the default off, and opting in explicit. A manager saying “I’d find notes useful for the project part — happy either way” gets a real answer. “I’ll just turn on notes” gets compliance.
• Either person can stop it in any week, without a reason. And say so plainly: no reason is a complete reason. This is the single most important part, because the value of a 1:1 depends on someone being able to change the terms in the moment they need to.
• Say when you stop, out loud. Silence isn’t legible. “I’m turning notes off for this bit” is four seconds and it’s the sentence that makes the rest of the arrangement believable.
• Know what you’re creating. A manager’s AI summary of a report’s career conversation is not personal notes; it’s a record about a named person that may outlive the relationship, especially on a company account with an admin console. Who can see my AI meeting notes covers where these actually live, and is it legal to record a meeting plus one-party vs two-party consent cover the shape of the rules, which vary by region. Your own HR policy is likely to be stricter than the law, and is the one that will actually bind you.

There’s also a middle setting people forget exists: you can have notes without keeping the audio. Meeting notes without recording the call walks through the difference between capturing a conversation and storing it — a smaller artifact is a genuinely smaller exposure, though it changes nothing about whether you should have mentioned it.

Where a live summary actually earns its place here

Being honest about the wedge: in most meetings a rolling summary buys back attention you’d otherwise have lost. In a 1:1 you didn’t lose any — you were there for all of it. So it’s buying something narrower, and it’s worth naming precisely.

It buys you the freedom not to write. The do I still need to take notes answer argues that once a tool handles the record, the only notes left are the ones that were never said out loud — what you committed to, what you disagreed with, the thing you thought of at minute nine. In a two-person conversation, writing even those is expensive, because there’s nowhere for your attention to go that isn’t visibly away from the other person. Which means you have to hold them instead, on top of listening and responding. A glance at a multi-resolution summary costs a second; reconstructing what someone said eleven minutes ago costs the thread.

The sharpest case is the one nobody plans for: the dense download. Every few months a 1:1 stops being a conversation and becomes eleven minutes of your manager explaining a reorg, a compensation change, a shift in priorities, or hard feedback about you. That’s a broadcast inside a 1:1, and you’re absorbing it under exactly the emotional load that makes working memory worst — which is precisely when you most need to respond well and are least able to hold what was said. A rolling summary sitting on screen is the difference between answering and nodding.

Canary shows that view while the conversation is happening, capturing your computer’s system audio with no bot in the call, no plugin, and no virtual audio device, on macOS, Windows, and Linux. It runs on your machine, which means it doesn’t need the other person’s platform or an admin’s permission to work — and that is exactly why the disclosure section above is the longest one on this page.

Where after-the-call notes are the better half

This page would be dishonest if it didn’t say that the biggest tooling win in 1:1s is a post-meeting one.

A 1:1 is a serial conversation — one instalment every week or two, and the whole thing runs on what did we say last time. Most start with four minutes of two people trying to remember, and end with commitments that neither writes down. Automatic notes and action-item detection genuinely fix that, and tools built around the after-artifact are good at it: Granola produces one of the best bot-free post-meeting documents in the category, and Circleback is strong on follow-ups. Better still for this specific job is a shared doc both people can see and correct, which is what Fellow and Notion are built for — two private summaries of the same 1:1 are two records nobody reconciles, while one shared page builds common ground during the meeting.

Two caveats before you rely on any of it. A fluent summary can be confidently wrong in ways that are hard to notice, which matters more here than in a status meeting, because a misremembered commitment in a career conversation is the kind of thing people fall out over — the meeting recall guide covers keeping a record you can actually trust. And a summary of a 1:1 is other people’s unguarded words with their name on them; the fact that it’s easy to forward is not a reason to.

The short version

If it’s a working 1:1 between peers, a client call, or the project half of a manager–report meeting: capture it, mention it once, and get the time back. If it’s the half where someone might say something hard: default to off, keep a pen for the four things only you know, and let the person with less power decide. Nothing about being back-to-back all day changes that trade — and a 1:1 that produces a great transcript and no honesty is a worse meeting than one that produces neither.