⌘
Runs on your device
Inference happens locally on your own CPU/GPU/NPU. No per-token costs, no rate limits, no outages.
Local LLM · zero cloud · private by design
Remember everything.
Share nothing.
frunteer quietly captures the people you meet and what matters — then transcribes and understands it all with a local LLM running on your own device. No cloud. No servers. No API calls. The most personal data imaginable never leaves your hardware.
Capture on-device
Reason locally
Works fully offline
The cloud can't leak what never leaves your device.
100%
of capture, transcription, reasoning & storage happens on your hardware
0
bytes of your conversations ever sent to any server
∞
offline — no internet required to capture or query your memory
How it works
Local-first, end to end.
Every step lives on your device. There is no cloud in this pipeline.
01
Capture on-device
A discreet companion records your day. The audio stays on your hardware from the first sample.
02
Run a local model
frunteer transcribes and reasons with an open-weight LLM — Llama, Gemma, Phi — running right on your CPU/GPU. No round-trips, no API keys.
03
Ask, privately
Search your entire life by meaning, instantly, offline. Your memory answers only to you.
The local LLM
Open models. Your hardware. Your call.
No black box in someone else's data center. You run the model, you see the weights, you keep the keys.
⟨⟩
Open-weight models
Built on inspectable open models. Bring your own — Llama, Gemma, Mistral, Phi. You choose what runs on your data.
⌁
Works fully offline
Capture, transcribe, reason, and query with no connection at all. Air-gap it and it still works.
⚿
Encrypted local vault
Your memory is encrypted at rest and you hold the only key. Lose the key, the data is noise.
⊘
No account, no telemetry
Nothing to sign up for. Nothing phones home. There's no server logging what you ask your own memory.
⌫
Delete = gone
Erase a memory and it's scrubbed from disk and model state. Permanent, verifiable, yours to command.
Privacy is the architecture
There is no cloud. Only your device.
frunteer processes the most personal data imaginable — your real conversations — with a model running on hardware you own. We can't read it. No one can. There is nowhere for it to leak to, because there is no "there."
- 100% on-device: capture → transcribe → reason → store
- No accounts, no servers, no telemetry, ever
- Open-weight models you control and inspect
- Encrypted locally; deletion is permanent
$ capture → whisper.cpp → local-llm → encrypted-vault ▸ all local
Carry your life into AI —
and keep it to yourself.
Join the frunteer waitlist for early access to private, on-device AI memory.