AI meeting notes without a bot

Bot capture vs local capture

There are two ways an AI notetaker can hear your meeting. A bot notetaker joins the call as a participant — you'll see it in the attendee list — and the meeting audio is streamed to the vendor's servers. A local notetaker records the sound on your own machine: your microphone on one channel and the meeting audio on the other, straight from your PC, with nothing added to the call.

Both can produce good notes. The difference shows up in three places: who knows AI is in the room, which meetings you can capture, and where your audio ends up. A growing number of tools now offer some form of bot-free capture, which is a good thing — but they vary a lot in which platforms it covers and whether your data still lands in their cloud afterwards.

Why people skip the bot

• It's quieter. No "Someone's Notetaker has joined" moment, no explaining the robot in the corner. You still tell people you're recording — that's on you, and it's the law in many places — but you do it like a person, not by proxy.
• It works in any app. Local capture records anything you can hear: Zoom, Teams, Meet, Webex, a webinar, a softphone, even a phone on speaker. There's no per-platform bot integration to break.
• It can keep your audio local. A bot almost always routes audio to the vendor's cloud. Local capture can keep the recording on your machine — which, for confidential calls, is the whole point. More on that in local vs cloud transcription.
• It captures meetings you don't host. You can't always add a bot to someone else's call. You can always record what reaches your own speakers (with the room's knowledge).

What to look for in a bot-free notetaker

"No bot" on its own doesn't tell you much any more. The questions that separate the options:

• Which platform is it on? Some bot-free modes are Mac-only right now. If you're on Windows, check that bot-free capture actually works on Windows, not just in the marketing copy.
• Where do the recording and transcript live? On your device, or back in the vendor's cloud after capture?
• Can you run it yourself? Open-source, self-hostable tools let you use your own AI keys and keep everything in your control.
• Does it capture both sides separately? Recording your mic and the meeting on separate channels is what makes speaker labelling reliable instead of guesswork.

How Earshot handles it

Earshot is built around bot-free, local capture on Windows 10 and 11. It records both sides of the conversation on separate channels right on your PC, so "Me" and "Them" are physically distinct audio and speaker attribution isn't a guess. It works in any app you can hear, and the recording and notes stay in a folder you control. From there you get a clean summary, decisions, suggested action items you approve, and Ask Earshot for questions across every meeting, answered with quoted receipts.

Two things make Earshot's version of "no bot" different from most: it's open source under the MIT licence, so you can read exactly what it records and where it sends things, and it's self-hostable — plug in your own transcription and AI keys (even a fully local model) and your audio never leaves your machine. Prefer zero setup? Earshot Plus runs the AI for US$9 a month, with audio processed and discarded, never stored.