Launching today
Ellis
AI notes for in-person meetings
138 followers
AI notes for in-person meetings
138 followers
Ellis is an AI notetaker for in-person meetings. Record your meeting, get a clean transcript with each speaker identified, then ask anything — what was decided, what you missed, how it went. No laptop. No extra hardware. Just your iPhone (or Apple Watch).
Recording doctor visits and therapy sessions puts this app in about the most sensitive bucket there is, so deleting the audio right after transcription is a smart move. The transcript still says everything the audio did, though. I saw Supabase in the stack, are transcripts scoped per user at the database level, or does the app layer do the gatekeeping?
Ellis
@vollos great point! Users are only able to access their own transcripts, which is enforced by the backend (not only in the client-side app). Curious about trying the app in one of these scenarios, or do you have more thoughts here?
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?
Ellis
@mia_qiao so excited that you ask :) So there's a multiple step process.
Users can record their voice during onboarding which works as a "voice enrollment."
When a meeting ends, the app displays several different cards, each with a quote and a "memorable" quote. The card that best matches the voice enrollment gets highlighted as "Best match"
The user can then assign themselves as the speaker, and add names to other attendees
When finalized the notes are updated with a new understanding of who you are in the group.
how does it handle cross-talk or overlapping speakers when multiple people talk at the same time during the meeting?
Ellis
@dnde8yg you get a full transcript where you can then easily identify speakers. By cross-talk there are occasional instances where it might identify two speakers as one. But the recent AssemblyAI model is surprisingly good and getting better.
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.
Ellis
@krekeltronics 100% agree.
Onepane
nice, everyone builds for zoom calls and forgets real meetings exist. curious how it does with a noisy room, that's where my notes always fall apart
Ellis
@ashmil_hussain give it a try and let me know! What type of noisy rooms are you thinking of? public places?
Recorded a quick team standup on my iPhone and the speaker identification was spot on, even with people overlapping. Asking it what got decided after felt like magic.
Ellis
@yusufgbelbkvy thanks for quick test and feedback! How many people were in the standup?
Runs quietly in the background and nails speaker labels without any setup. Wished I had this for those standup meetings where half my notes end up being nonsense.
Ellis
@zzet6v8x appreciate the review!