Ellis

AI notes for in-person meetings

Productivity

Artificial Intelligence

Audio

Visit WebsiteSee on Product HuntTwitter ⧉

Honestly it's rare by volume but it clusters right where it hurts, the fast back-and-forth when a decision actually gets made, so it feels worse than the raw percentage suggests. I wouldn't chase overlapped-speech separation, it's expensive and still flaky. I'd just surface it: flag the low-confidence diarization spans so I know which two or three lines to re-listen to, instead of trusting a transcript that looks clean. The silent winner-pick is the part that bit me, not the noise.

Everyone built for Zoom and forgot rooms exist. How does it handle four people around one table with a single phone mic?

In-person is the right wedge, every AI notetaker assumes a Zoom link exists. How do you handle speaker attribution in a noisy room without everyone wearing a mic? That's the failure mode that killed my voice-memo system for coffee meetings.

The in-person angle is what sets this apart, but recording therapy or doctor visits is also where the privacy question gets sharp. Does the audio and the diarization run on-device, or does the recording get uploaded to a server to transcribe and match against my saved voice profile? And since everyone else in the room never installed the app, is anything about their voice retained, or is it all local to my phone?

For me it comes up in fast 3+ person brainstorms and standups, almost never in 1:1s or sales calls where people take turns. I wouldn't chase true source separation, that's a research problem you don't want to own. The cheap win is honesty: when AssemblyAI hands back a low-confidence or overlapping stretch, drop a small 'crosstalk here' marker instead of a clean line, so I know to trust my own memory for that bit. A confidently wrong transcript is worse than one that admits a gap.

The noisy-room question has a nastier cousin: true overlap, two people talking at the same instant. Diarization picks one speaker per frame, so the quieter voice doesn't get mislabeled, its words just vanish, and nothing in the transcript tells you a sentence went missing. That's harder to catch than plain noise because the output still looks clean. When people talk over each other in a fast brainstorm, does Ellis mark the overlap or just pick a winner?

The decision to make it work on just an iPhone or Apple Watch is genuinely clever. No extra hardware means it'll actually get used in real meetings instead of sitting in a drawer.

How does the speaker identification actually work in practice, especially when people are talking over each other in a real meeting?

How well does the speaker identification work in a noisy cafe or group setting with people talking over each other?

How well does the speaker identification actually work when people are talking over each other or interrupting? That's usually where these tools fall apart for me.

Curious how it handles crosstalk or people talking over each other in a noisy room. Does the speaker identification still hold up, or does it get messy fast?

finally tried ellis at a coffee chat and the speaker labeling actually nailed it, even with overlapping talk. liked that i could just leave my phone on the table and forget about it

The speaker identification works surprisingly well even in a noisy coffee shop, and being able to ask follow up questions about a meeting I walked out of feels like a real superpower.

the fact that it runs straight off the iPhone and Apple Watch with no extra hardware is such a thoughtful move for in-person meetings.

How well does it handle overlapping speakers or side conversations in a noisy room, and is the transcription actually reliable enough for something like a legal or HR meeting?

In-person notes are a different trust problem than meeting-bot notes. The useful detail is speaker correction and ownership after the recording, because the transcript becomes part memory, part operating record. Getting that boundary right matters.

The fact that it runs straight from your iPhone or Apple Watch without any extra hardware is genuinely clever, removes so much friction from actually capturing in-person conversations.

The speaker identification without cloud processing feels really thoughtful, like the team actually thought through what people want during a meeting instead of just chasing the AI hype.

The “AI Notes for In-Person Meetings” angle is interesting, especially for teams that still do design reviews, sales conversations, or planning sessions around a table. How does Ellis know who said what in a room? Is speaker separation part of the flow, or is it more focused on capturing clean notes and action items after the conversation?