frequently asked questions

AvenBox runs a full AI pipeline locally: speech-to-text, large language model inference, text-to-speech, and tool execution — all without an internet connection. Think of it as a private AI assistant that never phones home.

No. AvenBox is designed to work completely offline. Wi-Fi is optional — used only for software updates and when you explicitly want to reach the internet. All AI inference happens on-device.

Those devices route everything through cloud servers. AvenBox processes everything locally on a custom AI engine. Your conversations, data, and commands never leave the device. Plus, it has real battery life, real storage, and real cooling — not a tech demo.

AvenBox operates in three modes: Solo mode (fully offline, handles simple tasks locally), PC Director mode (delegates heavy tasks to your own PC running a bigger model), and Privacy-Safe Cloud mode (sanitizes your prompts before sending to cloud AI — strips names, numbers, and sensitive data so the cloud never sees your private information).

Both share the same chassis, software, and capabilities. Sovereign has 8GB unified memory (runs 9B parameter models). Sovereign Pro has 16GB (runs larger 13B+ models with longer context). Same AvenCore AI Engine, same silent cooling, same build quality.

AvenOS is our custom Linux-based operating system designed for sovereign AI. It includes a privacy-first routing kernel that intercepts system intents and routes them to local models first, falling back to cloud only when explicitly permitted by the user.

First 500 Sovereign units ship Fall 2026. Waitlist position determines your priority in the queue.

All AvenBox units come with a 30-day no-questions-asked return policy. Full refund, no hassle.

Yes. AvenBox includes full SSH access and a model management CLI. Load any GGUF-compatible model that fits in memory. The Developer Pack will include pre-configured toolchains for popular model formats.

There is no telemetry, no analytics, no cloud sync by default. AvenOS runs an air-gapped inference pipeline. Network access is off by default and requires explicit user opt-in per feature. The device has no camera and no always-on microphone — mic activation is push-to-talk only. When you do use cloud AI, AvenBox sanitizes your prompts first — stripping personal information before it leaves the device.

Yes. AvenBox implements NIST-standardized post-quantum cryptography (ML-KEM + ML-DSA) for all device communications. Your data is protected against both current threats and future quantum computer attacks. No other consumer AI device offers this.

Updates are delivered via Wi-Fi when connected. They are optional and user-initiated — AvenBox will never auto-update without your permission. You can pin to any OS version indefinitely.