Project Mnemosyne 发布：耳塞加智能戒指，用语音指挥 Codex 和 Claude
一句话说清楚：有人在给 Codex、Claude 这些智能体做语音硬件了，耳塞加戒指，错误率比 AirPods 低很多，还能耳语输入。如果你天天开着 agent 干活，这套交互思路很值得看看，毕竟打字管理一堆 agent 确实别扭。
Dario Amodei 认为很快 AI 将端到端完成软件工程师的大部分工作，人类转为监督角色。针对这一场景，@antimattr_one 推出 Project Mnemosyne，一套 AI 耳机加智能戒指的组合，用于口述想法、捕捉灵感，并用语音操控 Codex、Claude 等智能体。据公司自测，其词错误率从 AirPods 的 30% 以上降到约 5%。产品还提供 whisper mode（耳语模式），解决在图书馆或共享办公空间不便大声说话的问题。作者判断语音将成为人机协作智能体的主要交互方式，而对应的硬件目前几乎空白。
现有交互方式的问题在于，如果工程师要管理多个自主运行的智能体，在聊天框里打字反而是低效的入口。Mnemosyne 把入口移到可穿戴设备上，让随时随地的语音输入和指令成为可能。对开发者而言，这意味着编码智能体的操作界面可能从 IDE 和终端之外再长出一条语音硬件赛道，目前尚无成熟竞品占据这一位置。
Dario Amodei thinks AI will soon handle most of a software engineer’s work end to end, with humans moving up into oversight.
If that happens, typing into a chat box is a strange way to manage your agents.
@antimattr_one just launched Project Mnemosyne, AI earphones plus a smart ring to dictate, capture thoughts and steer agents like Codex and Claude by voice. In the company’s own tests, word error rate dropped from over 30% on AirPods to around 5%.
What’s interesting is the whisper mode. Nobody wants to talk loudly to their agent in a library or a coworking space.
I think voice becomes the main way we work with agents, and the hardware for it barely exists yet.
This report is categorized under AITOP's "Products" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Project Mnemosyne, Claude, Codex, voice interaction.
Source attribution: curated from kimmonismus (@kimmonismus).
