v0.2 · live meeting copilot

The answer while they are still talking.

Umbra sits beside your call, hears both sides, and puts the sentence you need on screen in about a second. When the call ends, the recap and the follow-up email are already written.

  • ✓ No account, no sign-in, no bot in the meeting
  • ✓ Your own keys — or none at all
  • ✓ Follow-up drafted before you hang up
umbra Sales call ▾
them Question detected · 0.4 s ago

$18 a seat at 40, annual — and credit the pilot against the first invoice.

Why: the volume tier starts at 25 seats, so 40 lands one band down.

480 ms first word gpt-4.1-mini copy
Ask anything…
Interface preview · answers, then the follow-up
Three seconds to learn, every call

In the room, not three hours later.

No bot joins the call. No one else hears about it. Umbra listens to the audio already playing on your machine, answers while the question is still open, and closes the loop before you stand up.

  1. 1

    Start listening

    ⌘⇧L opens a channel to the system audio and your microphone, separately, so “them” and “you” stay distinct. A listening badge sits in the pill while it runs.

  2. 2

    Read the answer

    ⌘⇧A answers what was just asked. Or turn auto-answer on and the sentence is already there when the question lands — one line you can say out loud.

  3. 3

    Leave with the follow-up

    Decisions, owners and dates are captured as they are said. The recap and a ready-to-send email are waiting in the dashboard when the call ends.

What is actually in the build

Small, sharp, and honest about it.

Answers in the room

One line, phrased to be said out loud, with the reason underneath. Latency to first word is measured and shown, so you know what you are trusting.

Decisions, owners, dates

Every call becomes a meeting file: transcript, answers, notes. Search across all of them from the dashboard instead of scrolling a chat thread.

The follow-up, already written

A recap plus a sendable email, drafted from what was actually said. Copy it, edit the two words that are wrong, send it before the next meeting starts.

When the answer is on screen

⌘⇧R freezes the screen and lets you drag a box around a deck, a ticket or an error. Only that box is sent, cropped at native resolution.

Your models, or ours

OpenAI, Anthropic, Google, Groq, or Ollama on your own machine — switch provider and model in two clicks. A test button reports the real error instead of a shrug.

Private by construct

No account, no telemetry, no sync. Keys, transcripts and notes sit in one folder you can open and delete. Nothing leaves the machine except the request to the provider you chose.

Works with no network

Offline mode refuses every request before a socket opens. Pair it with Ollama and a local model and the whole loop — transcription aside — stays on the machine.

Out of your own screen share

The panel sets content protection, so when you share your screen the usual conferencing apps capture a blank area instead of your notes. Best effort — measured by the app, not promised by marketing.

Zero-key local engine

On first launch, with nothing configured, Umbra still answers by pulling the strongest lines out of what was said. It says that is what it did, every time.

The part most tools bury

Nothing to log into.
Nothing to leak.

There is no Umbra account, because there is no Umbra server holding your data. Settings, meetings and model keys all live in one folder on the machine in front of you.

If you use a hosted model, the audio transcript and the region you selected go to that provider — that is the deal you are making with your own key. If you use the included model, they go through our proxy instead, counted by the hour and not stored. Offline mode refuses both.

Read the honest limits →
  • No account Nothing to sign up for, nothing to forget. Download, open, done.
  • No bot in the meeting It reads audio already on your machine. Nobody is admitted to the call.
  • No telemetry The app makes no network call except to the provider you configured.
  • No cloud storage One JSON file per meeting, atomic writes, a delete button that actually deletes.
Download

Open it and it already works.

No installer wizard to click through, no account to create, no key to paste before your first answer. Add a hosted model later if you want longer answers.

macOS · Apple silicon Umbra-0.2.0-arm64.dmg 119 MB · macOS 12 or later · unsigned Download .dmg
Windows · x64 Not published yet The NSIS installer is configured and builds, but no signed Windows artifact has been produced or tested on real hardware. Rather than hand you an untested .exe, this one waits. Build it from source

Install on macOS

  1. 1
    Open the .dmg and drag Umbra to Applications. One window, one drag. There is no installer to run.
  2. 2
    Right-click Umbra in Applications → Open. The build is unsigned, so Gatekeeper blocks a double-click the first time. This is expected, and it is a one-time step. If macOS refuses anyway, run the command below once.
  3. 3
    Allow Microphone and Screen Recording when asked. System Settings → Privacy & Security. Microphone is for the “you” channel; Screen Recording is for region capture and system audio. Restart after granting them.
  4. 4
    Press ⌘⇧L on your next call. That is the whole setup. Add a model key in Settings when you want more than the local engine.

Build it yourself

Node 20 or later, then three commands. The same three produce a macOS or Windows installer.

npm install
npm run smoke      # boots it, runs one ask, exits non-zero on failure
npm run dist:mac   # or dist:win
Muscle memory

Five shortcuts, no menus.

⌘⇧SpaceShow or hide the panel
⌘⇧AAnswer what was just asked
⌘⇧LStart or stop listening
⌘⇧RSelect a region and answer it
⌘⇧HHide everything, now

On Windows, Ctrl replaces ⌘.

Licence

Free with your own keys.

Bring a model key, or bring nothing and use the local engine — either way there is no seat count, no trial clock, and no activation server to be offline from. Paying is for the convenience of not managing a key.

Your own model keyFree
Local engine, no keyFree
Included model + transcriptionPlanned
Account requiredNone

The included model is metered by meeting hours, because transcription is what costs money — not by answer. Checkout opens in your browser and the licence key is pasted into the app; the app itself never needs a login.

Read this before you rely on it

What Umbra does not do.

Every tool has edges. These are the ones that would annoy you if you found them later.

  • It does not decide who consented. Recording law differs by state and country, and two-party consent is common. Disclosing that you use a copilot is your call to make, in the room, before the call — not the app's.
  • It is not invisible to everything. Content protection is respected by Zoom, Meet, Teams and QuickTime. A capture tool that ignores it — or a phone camera — will still see the panel.
  • The local engine does not reason. It finds and quotes the strongest matching lines from what was said. It says so, every time. Add a model for a real answer.
  • Retrieval is lexical. BM25 plus trigrams over your own meeting files. No embeddings, no vector database, no downloads.
  • English-first prompts. Other languages work through your model and speech provider, but the built-in prompts are written in English.
  • Unsigned builds. No Apple Developer ID yet, so the first launch needs right-click → Open. Windows has not been run on real hardware.
Questions worth answering

Short answers.

Does a bot join my meeting?

No. Nothing is admitted to the call and no participant list changes. Umbra reads the system audio already playing on your machine, plus your microphone on a separate channel, so it knows who said what.

Do I need an API key to use it?

No. With no key, Umbra runs its local engine: instant, extractive, honest about being extractive. Add an OpenAI, Anthropic, Google or Groq key — or point it at Ollama — when you want a model that reasons.

Where do my meetings go?

One JSON file per meeting in ~/Library/Application Support/Umbra/data. Settings sit beside them. Delete the folder and everything is gone — that is the whole backup story.

Will the other people on the call know?

Only if you tell them. The panel keeps itself out of your screen share and stays out of the participant list, so nothing announces it. That makes disclosure your decision, and we think you should make it.

Can I use it in a job interview?

If the interviewer knows, yes — it is a good cue card. There is no mode for pretending you are not using it, and we are not adding one. For the interviewer's side, the Recruiting mode is built for probing depth and taking notes.

Do I have to log in on the website?

There is nothing to log into. Downloads are plain files, settings live in the app, and a licence key is pasted into the app rather than tied to an account here.