Every meeting, on the record. None of it on our servers.
Ledgeur transcribes your meetings and works out who said what — the speech model and the speaker model both run on your machine. Name a voice once and it is recognised in every meeting after that. No bot joins the call. No minutes to buy.
No sign-up needed to record. An account only adds sync and agent access.
The only thing I want to settle today is whether we ship the pricing change before or after the conference.
Before. If we wait we spend the whole conference explaining a price nobody can buy yet.
Then we need the billing migration done by Thursday. Sam, can you own that?
I can, but I want the rollback path reviewed first. I'll have something to look at tomorrow morning.
- Sam — billing migration, with a reviewed rollback path, by Thursday
- Ship the pricing change before the conference
GradeHackBenefilyBill100BrandmityChartelyClassify7ConformeryContextelyContract10DeckHelmDenial7DuebayEvent70FeedlarkFoundily
Frifti
GrannioInkRobinJob13JobPlumbLedgerageLedgeraryLedgerlerLugbirdMealaryMeteraryModelCharterPatent77PatientaryPerDiemWisePermitBirdPortRobinRenewBirdRigbirdRoofHelmRotaBayScouternScreen100SlopeifySourceorySpend7TableHelmThreadCampTrial0
Under HausVouchityWageCoach
WatermarkRemoverProWillThisHappenA record you own, not a subscription to your own conversations.
Every other AI notetaker is a pipe to somebody else's database. That is a design choice, and it is the one thing Ledgeur does differently.
The audio never leaves
Whisper runs in your browser. So does speaker separation. There is no upload step to trust us about — open the network tab and check.
It learns who people are
Ledgeur separates the voices in a recording, and once you have named one, it recognises that person in every meeting afterwards. The voice prints stay on your device.
Nobody joins your call, and nobody can lock it out
No bot in the participant list, no awkward pause while people ask what it is. Ledgeur listens to the tab, the way you would. Which also means no video platform can revoke its access, the way they can with a bot that has to be admitted as a guest.
Free is the whole product
Unlimited recording, transcription, speakers, notes and search — permanently, for nothing. You pay when you want the record shared across a team, or open to your AI agents.
It learns the voices in the room.
A transcript that says “um, right, so” for forty minutes is a wall. A transcript that says who said it is a record you can act on.
Ledgeur runs a speaker segmentation model over the audio to find where the voice changes, then turns each stretch of speech into a voice print and groups them. You get Speaker 1, Speaker 2, Speaker 3 — with the overlaps handled, because people talk over each other.
Rename Speaker 2 to Priya once. From then on, Ledgeur recognises Priya in every meeting she is in. The voice prints live on your device and are never synced, never uploaded, and never part of the paid tier — a voice print identifies a person even after the transcript is deleted, so it stays where it was made.
- 1The recording ends
Ledgeur finds the turns and gives each voice a print — a 256-number fingerprint of how that person sounds.
- 2You name one
Click “Speaker 2”, type “Priya”. The print is saved under that name, on this device only.
- 3Next Tuesday
Priya speaks. Her print matches. The transcript says Priya before you have read a line of it.
- 4It keeps learning
Each meeting refines her print as a running average, so a bad headset once does not undo ten good recordings.
Three steps, and none of them are “create an account”.
Capture
Share the meeting tab with its audio, or just your microphone. Or drag in a recording you already have — a voice memo, a Zoom export, an old interview. It is treated exactly like a live meeting.
Transcribe on your device
Whisper runs in the browser through WebGPU, or the CPU if there is no WebGPU. The first run downloads the model once; after that it is cached and works with the wifi off.
Read it, and act
Speakers separated, timestamps on every line, a summary with the decisions and action items pulled out. Edit it, export it, search it later.
What actually differs from a cloud notetaker.
Not a feature-count. These are architectural differences — the consequences of where the audio goes.
| Ledgeur | A hosted notetaker | |
|---|---|---|
| Where the audio goes | Nowhere. Transcribed in your browser. | Uploaded to the vendor's servers. |
| Who joins the call | Nobody. It captures the tab's audio. | A bot appears in the participant list. |
| If the video platform tightens bot access | Nothing changes. There is no bot to admit. | The product stops working on that platform. |
| Minutes per month | Unlimited — it is your CPU. | Capped, then metered. |
| Cost for one person | Free, permanently. | Per seat, after a trial. |
| If the company disappears | MIT source, local files. It keeps working. | Export before the lights go out. |
| Reading your data with an agent | An MCP endpoint you point Claude at. | Whatever the vendor's integrations allow. |
Your agent is guessing about your work. The answer was in a meeting.
Why the architecture is like that, what the customer actually objected to, which decision was quietly reversed: it was all said out loud, and none of it is in the documentation. That makes the meeting record the highest-context thing a company produces and the least reusable.
So Ledgeur opens it over the Model Context Protocol. An agent can list your meetings, search them, read a full transcript with speakers, and pull the open action items, over an open standard rather than one vendor’s private API.
Access runs as you: the token resolves to your session, so row-level security decides what the agent can see. It cannot read a meeting you could not.
list_meetingsBrowse the most recent meetings.search_meetingsFind a meeting by what it was called.get_meetingThe full transcript, speakers and notes for one meeting.list_tasksEvery action item, filtered by status.list_peopleEveryone named across your meetings.
Works with every browser-based meeting platform.
Coming from something else?
A macOS AI notepad that listens to your meeting audio and enhances your typed notes with an AI summary after the call.
A cloud transcription service that records meetings, generates live transcripts and AI summaries, and stores them in your Otter account.
A meeting assistant bot that joins calls via a calendar integration, records and transcribes them, and pushes notes to your CRM.
A free-for-individuals notetaker that records Zoom/Meet/Teams calls via a bot and produces highlights and summaries.
A meeting recorder that captures Zoom/Meet/Teams calls with a bot, timestamps highlights, and shares clips.
A revenue-team meeting assistant with transcription, conversation intelligence and CRM sync.
Built for every kind of meeting.
Or take the template and do it by hand.
The six templates the app runs on, published as headings you can paste into a blank document.
Your meetings, your machine, your record.
The whole product is free for one person, permanently — not a trial, not a tier with the good parts removed. Pay $12 a month per person only when you want the record shared across a team and readable by your agents.