Skip to main content

Proof

What runs on
your machine.

Kajo Voice runs speech recognition, translation, summarization, and chat on your Mac, Windows, or Linux computer. Imported audio is transcribed by an on-device model (Whisper large-v3 turbo) into a local encrypted library; a single on-device Gemma 4 model then translates, summarizes, and answers questions — all on-device. There is no cloud-processing path: your audio, transcripts, and library never upload.

What never leaves your laptop

Every stage of the loop runs on the machine the audio is already on.

Your machine

No cloud-processing path

  1. Drop in

    Audio or video, from your Mac, Windows, or Linux machine.

  2. Transcribe

    Whisper large-v3 turbo, running on that same machine.

  3. File

    Into an encrypted kajo.db — your local library.

  4. Ask

    Translation, summaries, and cited chat — one on-device Gemma 4 model, over that same library.

The default install never crosses the network with audio. There is no cloud-processing path — transcription, translation, summaries, and cited chat all run on your machine.

Full Trust Center → · Security & encryption →

The models that run on your machine

Every step of the loop has an on-device model. These are the exact engines Kajo ships — verifiable in the packaged app and pinned in CI.

Task Model License
Transcription Whisper large-v3 turbo 98 languages, batch transcription, Mac / Windows / Linux. MIT
Translation + summaries + chat Local LLM — Gemma 4 One on-device model translates transcript text, generates summaries, and answers chat questions locally; Gemma 4 12B is Apache-2.0. There is no cloud option and no separate translation model. Apache-2.0
Library search (embeddings) Qwen3 Embedding 0.6B Dense vectors for hybrid search across your archive. Apache-2.0
Search reranking BGE Reranker v2 m3 Cross-encoder reranker for cited chat retrieval. Apache-2.0

We track each model's license so you can vet what ships before deploying on a regulated machine. Full attribution lives in the packaged app's third-party notices.

Three pillars behind the diagram

  1. Local-first / private by default

    Your audio never leaves your laptop.

    Audio is transcribed on-device — there is no cloud-processing path. Private by architecture.

  2. Cross-platform, including Linux

    Mac, Windows, and Linux — one product, three operating systems.

    Native installers for all three. Linux is a first-class target, not a port — the only local audio knowledge base that takes it seriously.

  3. Cited chat

    Ask questions across your archive — get cited answers.

    Chat with your entire library using a local model. Every answer links back to the exact moment in the recording — no copy/paste.

Frequently asked questions

Does Kajo upload my audio to transcribe?

No. Transcription runs locally — and so do translation, summaries, and chat. There is no cloud-processing path; Kajo never uploads your audio or transcript text.

What leaves my laptop when I use Kajo?

Nothing from transcription, translation, summaries, or in-app chat — those all run on-device. License validation may use the network. Product analytics are opt-in; crash reports are opt-out and never include audio or transcript content.

How is the database protected?

kajo.db is encrypted at rest with a key wrapped by your OS keychain. Production builds refuse to start if the keychain cannot protect that key.

Join the waitlist