Voice control coding agents on your machine via smartwatch / CarPlay
TLDR: got tired of my desk, so now I touch grass while coding from my watch. Ever since the 2021 GH Copilot beta saved me from dual carpal tunnel, I figured AI's main benefit in coding would be cutting typing while keeping productivity. That's turned out true; no one writes their own code lately. Why wouldn’t you have AI do the first draft? Why not your updates too? I expected phone coding to rise next, but tiny keyboards wreck your hands even faster. To put this into practice, I started leaving my Windows desktop on 24/7 and SSHing in from my phone, yet the problem just transformed:…
What it does
In the maker’s words, at launch
TLDR: got tired of my desk, so now I touch grass while coding from my watch. Ever since the 2021 GH Copilot beta saved me from dual carpal tunnel, I figured AI's main benefit in coding would be cutting typing while keeping productivity. That's turned out true; no one writes their own code lately. Why wouldn’t you have AI do the first draft? Why not your updates too? I expected phone coding to rise next, but tiny keyboards wreck your hands even faster. To put this into practice, I started leaving my Windows desktop on 24/7 and SSHing in from my phone, yet the problem just transformed: now I had finger pain typing long prompts to Claude Code. I also commute 2 hours a day between SF and PayPal HQ, where I work as an AI Specialist / Sr SWE overseeing our AI coding-agent tooling. During that commute I couldn't touch my machine's work at all. So the app was initially just voice-activated coding for CarPlay, so I could dispatch and talk to agents entirely by voice while driving, like TTS texting. While building it, a friend told me about his older brother, who dispatches Claude Code by voice with a lav mic on his collar. Looks a bit strange, but you can instantly see how it works: he walks around outside and voice codes. That's when it clicked that I could do a native watchOS app. I'd owned an Apple Watch for years and barely used it; it's now my favorite use case. How it works: you speak from your watch, phone, or CarPlay, and the app opens SSH sessions into your own machines (any number), where your coding agent of choice runs. A go-between agent interprets what you say, dispatches the work, and reads results back when done. It has context compression and message queuing so you can keep talking while sessions run. Backend: two options. Use my cloud service for chat history/configs, or self-host. The backend is a Java Spring Boot app shipping as a single executable JAR with an embedded web server and H2 database, so you just need Java 21. No separate database to configure; it creates its data file under ~/.dashvox on first run. You don't need an account, and your code could avoid leaving your machine. Self-hosting also supports VPNs like Tailscale between phone and machine, which is how I run mine. The only thing that always touches my cloud is push notifications. APNs requires an auth key that must stay server-side, so the backend holds it; your device registers its APNs token on open so it knows where to deliver. That's the only data routing through me; your sessions and code don't. Cost: the app is free and I want to keep it that way. COGS are near zero since you bring your own machines and keys, so you only pay your existing providers. There's an optional donations page in settings; at most I'd add a small one-time purchase someday, never a subscription. You connect your machines, an LLM API key (~$25-50 lasts a while), and optionally a Twilio key for texting; the go-between offloads most work to your coding CLI. I tried configuring Siri, but Siri is honestly terrible, so I added Twilio: you can text the agent through Messages (already Siri-native). It's built for technical users. I’d steer non technical users to regular coding agents first; the value only lands once you're already paying for one, sick of waiting at your desk, and want outside. So this is for engineers and founders. The link is above; feedback and suggestions welcome. iOS is about as far as I can take it (iPhone / Watch / CarPlay). Next I'm exploring Meta glasses, though their voice API is restrictive. Android is nearly done (I built it first, I'm mainly a Java dev), but the Play Store needs 12 beta testers and I'm not sure I know that many Android users; if you'd want to be one, reach out. Ultimately I just want people to get outside more without choosing between that and shipping their own work. I haven't found anything else that does quite this, and it's hands down the best way I've found to remotely control a computer.
Does the same job
all alternatives →
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, June 2026
the whole month →
Fundraisly▲1,544AI fundraising agent that finds investors and books meetings
AI · Jun 2026 · fundraisly.com
- H6Homebrew 6.0.0▲1,481
Today, I’m proud to announce Homebrew 6.0.0. The most significant changes since 5.1.0 are a new tap trust security mechanism, the new faster, smaller, default internal Homebrew JSON API, sandboxing on Linux, better defaults informed by our user survey, many brew bundle improvements, improved performance and initial support for macOS 27 (Golden Gate). Happy to discuss any questions here!
Dev tools · Jun 2026 · brew.sh
- PU
hope you enjoy
Life & fun · Jun 2026 · vorpus.github.io
- IM
Life & fun · Jun 2026 · hackernewstrends.com