Cloud personality AI, voice, avatars, firmware, and fleet ops ā built as one integrated platform. Use it as a service or partner with us to ship your own.
A managed AI runtime for bots with their own voice, mood, and memory. Streaming chat, tool-calling, and persistent personalities.
XTTS voice cloning, faster-whisper STT, and a managed TTS fleet ā the same voices on the web, on phones, and on firmware devices.
A SoulBox runtime for ESP32-S3 boards. On-device wake words, low-latency audio, and OTA fleet updates from the cloud.
Generate avatars, talking-head video, and AI imagery on managed GPU fleets. Per-bot avatars, on-demand portraits, and image gen.
A control plane for multi-tenant device, voice, image, and inference fleets. Observe latency, scale instances, and manage providers.
OAuth + SSO, multi-tenant accounts, Stripe billing, and roles. Everything you need to ship a real multi-tenant product.
Every layer of the SoulBox platform is tuned for sub-second voice round trips ā cloud, codec, and firmware moving as one.
Device wakes on a custom on-device wake word ā no server round-trip required.
Stream audio to the SoulBox cloud, transcribed live with faster-whisper.
Routed through the right model with persistent personality, mood, and memory.
Streaming XTTS playback with avatar lip-sync ā back to your device in under a second.
Bring your own hardware, your own voices, or your own brand. We'll handle the cloud, the firmware, and the fleet underneath.
