The Zoom Hack That Buys Back Your Brain

Welcome to Briefly AI, a podcast by Harry Sharman, created by AI and voiced by an AI synthesis of Harry Sharman. A collaboration between Harry and a machine — the machine doesn't have anywhere else to be either.

Right. There's a small, slightly rebellious trick buried in this week's news, and I want to tell you about it — because it's genuinely useful, costs nothing, and also says something quietly important about the moment we're in.

Here's the thing. If you're in back-to-back meetings, you've almost certainly noticed that AI transcription has gone from opt-in novelty to just... what happens now. Zoom AI Companion, Microsoft Copilot in Teams, Otter.ai running in someone's browser — the meeting gets recorded, it gets transcribed, it gets summarised into a neat little five-bullet digest that sits in your inbox. And everyone quietly assumes everyone else is reading it.

The TechCrunch piece this week asked a good question about all of this: if every meeting gets transcribed and summarised, who's actually reading any of it?

The answer, in most organisations, is: almost nobody. But the recording keeps happening anyway.

So here's the thing worth knowing. You can stop it. Or at least, you can stop it from happening to you, in specific situations where you'd actually prefer to think out loud without generating a permanent searchable record.

This isn't about paranoia. It's about something more practical: when you know you're being transcribed, you edit yourself. You round off the rough edges. You don't float the half-formed idea. You don't say "I'm not sure this is right, but..." — because hedged thinking looks bad in a transcript. So you perform certainty you don't have. And the meeting gets worse.

So, the technique. And it's deceptively simple.

If you're the meeting organiser in Zoom, you can turn off AI Companion for a specific session without changing your account settings. Before the call, go into the meeting settings, find the AI Companion section, and disable the transcript for that meeting. Done. Nothing is recorded or summarised. In Teams, it's similar — when you start a meeting recording, you choose whether transcription runs alongside it, and you can start recording without transcription active.

But here's the more interesting version of this — what to do if you're not the organiser, and AI transcription is already running in someone else's meeting.

You can simply say so. You can say, "Hey, before we get into this — is the AI Companion running? I'd prefer this one stays off the record." In most organisations, that's a completely reasonable request, and most hosts will turn it off without blinking. The trick is knowing it's available as an option, and being willing to name it.

Let me give you a concrete worked example.

Say you're a senior manager and you're having a one-on-one with a direct report who's struggling. Not performing badly — struggling. There's nuance there. You want to have a real conversation: exploratory, honest, maybe a little uncomfortable. The kind where you're working out what the right support looks like in real time.

That meeting should not be transcribed. Not because anything inappropriate is happening — but because the transcript would create a sanitised, out-of-context record of a conversation that was specifically designed to be messy and human. The five-bullet summary would miss the point entirely, and it would sit somewhere in a system, searchable, attributable, in perpetuity.

So: before the call, disable AI Companion. If it's in Teams, don't start the transcript. Run the meeting as a meeting — two people talking, no record.

Then, if there's anything that needs to be documented — actions, decisions, follow-ups — you write that yourself, deliberately, after. You get to choose what the record says. That's actually how professional judgment used to work, before transcription became the default.

Now, where does this break down?

A few places. If your organisation has a policy that all meetings must be recorded — compliance, legal, regulated industries — this won't fly, and you shouldn't try to circumvent it. The tool is useful; the exemptions are real. Also, the transcript is genuinely useful for some meetings. Long project kick-offs, technical deep-dives, anything where "what did we actually decide" is going to matter later — transcription earns its place there. I'm not arguing to switch it off wholesale.

And there's a version of this that tips over into something slightly strange: people who are so transcript-averse that they're essentially asking colleagues to fly blind on every call. That's not discretion, that's just opacity. The useful version is intentional — knowing when the record helps, and when it gets in the way.

The broader point — and Harry's written about this in the context of cognitive offloading — is that handing your thinking over to an automated system has a cost that doesn't show up on the invoice. When AI transcribes and summarises everything, you start to assume the summary is the meeting. You stop listening as carefully. You stop retaining. And eventually, you need the transcript because you've outsourced the memory.

Choosing not to transcribe a meeting, deliberately, is a small act of claiming that cognitive work back. It forces you to actually be present, actually synthesise, actually decide what matters. That's not inefficiency. That's just thinking.

So the one crisp thing you can do tomorrow morning: check the default transcription setting on your next one-on-one or sensitive conversation, and make the decision consciously — record or don't — rather than letting the default decide for you.

That's Briefly AI. Harry talks to it more than he talks to most people, and honestly, it's a better listener.