Skip to content

FAQ

The app runs on your hardware. The only running costs are ones you opt into: cloud AI usage on your own key (or free tiers: Groq’s is generous), and ElevenLabs voice if you choose it. A local configuration (Ollama + Pocket voice) has zero marginal cost.

Conversation can: local mind, local voice, local speech recognition. See Privacy & your data. Web answers, weather, news, and account sync naturally need the network.

No. That’s the point. Claws are earned by spending time in conversation, only.

Are my conversations used to train anything?

Section titled “Are my conversations used to train anything?”

Not by Unclaw. Conversations go to the AI provider you chose, under that provider’s terms. Pick a provider (or a local model) whose policy you’re comfortable with. Unclaw’s own sync stores your history for you, to give your companions memory across your machines.

macOS (Apple Silicon), Windows, and Linux. See Install for what each needs. A phone companion (a video call with your companion, streamed from your own computer) is designed and on the roadmap.

Why does she download so much on first run?

Section titled “Why does she download so much on first run?”

The character engine and voices are real production assets: the same class of technology as a AAA game character. Shipping them separately from the app keeps updates small afterward.

Not currently: the stage belongs to one companion at a time. The roster switch is instant, though.

Tell us at unclaw.io. These docs grow with the app.