Your meetings,
remembered.
Local recording, speaker-labeled transcripts, a live assist overlay, and hands-free dictation — with a memory that recognizes people and answers questions across every past meeting. No cloud, no account.
$ curl -fsSL https://jiva.works/install.sh | bash
$ brew install --cask venugopalarjun/jiva/jiva
xattr -dr com.apple.quarantine /Applications/Jiva.app once in Terminal before opening it.
Everything, on-device
The whole pipeline runs locally — your audio never touches a server.
Live transcription
Both sides of the call, transcribed on-device as you talk.
Speaker memory
Labels who said what — and recognizes the same voice across every meeting, so names stick instead of “Speaker 1.”
Live assist overlay
A private glass overlay over any app: a running summary, one-tap Assist for what to say next, and a box to ask your past meetings — invisible to screen share.
Hold-to-talk dictation
Hold a key, speak anywhere, and it types for you.
Action items, automatically
Decisions and to-dos are pulled out for you — each linked to the exact moment it was said.
Auto-record
Starts the moment a meeting begins, stops when it ends.
Cited, not invented
Every summary, decision, and answer links back to who said it and when — tap a citation to jump to that moment.
Knowledge graph
See how a topic evolved across all your meetings.
Up and running in three steps
No Python, no setup — the app is fully self-contained.
Run one command
Paste the install command above into Terminal. It downloads, verifies, installs, and opens Jiva for you.
Grant permissions
A Welcome window walks you through microphone, system audio, and dictation access.
Models download
Dictation and transcription models fetch automatically — then record a meeting or hold to dictate.