开发者发布 Amisoria——一款 iPhone 应用，可为运行在自家 Mac 上的 OpenClaw 网关智能体添加 3D 角色形象、神经语音、口型同步、免提对话与插话功能，支持 20 种语言，基础功能免费且部分开源。
Posting with the mods' OK (asked via modmail first). I'm the developer. I wanted to talk to my OpenClaw agent the way you'd talk to a person, so I built Amisoria: an iPhone app that is only the face and voice of your own gateway. Nothing runs on my side. How it works - Your gateway stays on your own Mac (mine is a Mac mini). The app connects over your own Tailscale network with wss + an Ed25519 device key you approve once with openclaw devices approve. - Voice is your gateway's tts.convert (Microsoft neural voices, free). A small bridge on the Mac serves the audio file to the phone because the gateway has no download endpoint. - Lip-sync is text-driven: hanzi → pinyin → visemes, English through TalkingHead's own lipsync, spread over the clip length. No acoustic guessing, so it works in any of the 20 languages. - Hands-free: talk, pause, it sends; it listens again after the reply. Barge-in (interrupt her mid-sentence) uses iOS voice processing with her playback routed through the same audio engine so echo cancellation gets an exact reference — this took most of the last two releases to get right. - Works with whatever model your gateway is set to: GPT, Claude, Gemini, DeepSeek, Grok, Qwen, or a local Ollama model. - Verified against OpenClaw 2026.9.8 (the 9.7 stream change to data.delta is handled). The free tier is the whole thing: full conversation with your own agent, the 3D character, neural voice, hands-free and barge-in, 20 languages, no account, no ads, no time limit. There is a paid tier for a few extras (more characters, switching models from the phone), but you don't need it. What's open source: the host setup script, the bridge, install guide (EN / 繁中), persona rules and the character assets (Blender + MPFB, built from scratch — no licensed avatars): https://github.com/amisoria/amisoria App Store (iPhone, free): https://apps.apple.com/app/amisoria/id6805385809 There's an on-device demo mode if you want to see it before setting up a gateway. Happy to answer anything about the gateway protocol, the voice loop, or the character pipeline.
A developer built OpenClaw Voice, a dedicated voice interface using Home Assistant Voice PE and OpenAI Realtime, enabling sub-second smart home control, memory-based queries, phone calling, and long-running tasks with voice feedback. The entire stack is open-sourced.
OpenClaw has launched native mobile apps for iOS and Android, allowing users to run AI agents on the go with features like channels, tasks, and replies.
PhoneClaw is an open-source project that runs the AI Agent entirely locally on the iPhone, based on models like Gemma 4 and MiniCPM-V, no internet or data upload needed. It supports on-device operations such as voice, calendar, health data, ensuring privacy and fast response.
OpenClaw launched its official iPhone app, enabling users to pair with their private OpenClaw Gateway for chat, voice, approvals, and device-aware automation directly from their mobile device.