LOCAL-FIRST · LINUX-FIRST · YOURS
The answer, before
you're done being asked.
Sideglance listens to your meetings and interviews, transcribes them on your machine, and floats glanceable AI answers the moment a question lands. Your transcripts never leave your computer.
FREE FOREVER LOCALLY · NO ACCOUNT REQUIRED
HOW IT WORKS
CAPTURE
Both audio sides — your mic and the meeting — stream through on-device voice detection and whisper.cpp. Nothing uploads.
DETECT
When the transcript contains a real question, context assembles: recent conversation, your screen, your documents.
GLANCE
A headline and three bullets stream into the overlay in about a second. Read it mid-sentence; nobody knows.
WHAT IT DOES
Hears both sides
Mic and system audio, transcribed on-device in real time. You and They, color-coded, ~400ms behind the words.
Knows a question when it hears one
Question detection watches the conversation and triggers an answer only when one is actually needed.
Answers from your documents
Feed it your resume, product docs, meeting notes. Relevant passages are retrieved per question — locally.
Invisible overlay
Transparent, always-on-top, glanceable. Headline plus three bullets — read while the other person is still talking.
Remembers every meeting
Searchable history, one-click summaries with action items, Markdown export. All in a SQLite file you own.
Cloud only when you choose
Deepgram Nova-3 or Groq Whisper per meeting, falling back to local automatically. Never required.
THE INVARIANT
Your transcripts, your documents, your meetings — stored on your machine, and nowhere else.
Every competitor pipes your conversations through their cloud. Sideglance keeps speech-to-text, search, and storage on-device. Paid tiers proxy only the moment of a question — the prompt in flight, never persisted, never mined. The archive of what you said in every meeting you've ever had lives in one SQLite file that belongs to you.
PRICING
Local is free. Convenience is $12.
Free
Local forever. Bring your own keys.
- ·Unlimited local transcription (whisper.cpp)
- ·Unlimited AI answers with your own API keys
- ·Knowledge base, history, summaries — all local
- ·Every privacy guarantee, no account needed
Pro
No keys, no setup. We carry the API bill.
- ·Everything in Free
- ·Managed AI — zero API-key setup
- ·Deepgram Nova-3 cloud transcription
- ·30 managed hours / month
- ·Smart-model meeting summaries
- ·Email support
Power
Best models, serious volume.
- ·Everything in Pro
- ·GPT-4o / Claude-class answers & summaries
- ·100 managed hours / month
- ·Priority support
DOWNLOAD
Ubuntu today. More soon.
.deb and AppImage builds for Ubuntu 22.04+ on X11. Windows and macOS are on the roadmap — leave your email at launch and we'll tell you the day they ship.
QUESTIONS
Is my audio sent to your servers?+
No. Transcription runs on your machine with whisper.cpp. On paid tiers, only the in-flight prompt for a detected question touches our proxy — never stored, never your transcript archive.
What platforms are supported?+
Ubuntu Linux (X11) today — the only meeting copilot that treats Linux as first-class. Windows and macOS are on the roadmap.
Do I need an account?+
No. The Free tier needs no account, no sign-in, nothing. Accounts exist only to carry managed API usage on paid tiers.
Which AI models does it use?+
Free: bring your own Gemini, OpenAI, or Groq key. Pro: managed fast models. Power: GPT-4o / Claude-class for answers and summaries.
Will it show up in screen shares?+
Share a window instead of your full screen and the overlay stays invisible to others. Full-screen shares will include it — same as every tool in this category.