Sente

Hi Product Hunt 👋 — I'm Yuki, building Sente in Tokyo. Sente started from a selfish problem: I was paying for several AI subscriptions and still typing everything. So I built one agent that runs in my terminal (and on my phone) that I can talk to — and that answers back in my own voice. Three things make it different: 🎙️ Voice, both ways. Register your voice once (one sentence, ~10 seconds). After that you speak your request and hear the reply in your voice. Not a generic TTS voice — yours. 🔀 404 models on one plan. Claude, GPT, Gemini, DeepSeek and others through one account. Switch per task without rewriting your code — it's an OpenAI-compatible and Anthropic-Messages-compatible gateway underneath. 📖 Open source (MIT). github.com/yukihamada/sente — read the installer before you run it. `te privacy` prints exactly what leaves your machine. Who it's for: people who already pay for more than one AI subscription, and people who think faster out loud than they type. Honest limits, so you don't waste your time: - It's not "unlimited." You get a monthly credit allowance; heavy use means topping up. - Voice cloning requires your own consent and your own recording. We don't clone anyone else's voice. - The latency numbers on the site are measured, not estimated — but they're our measurements, from Tokyo. What I'd love from you: try one real task — a reply email, a summary, a small bug fix — and tell me where it broke. I'll be here all day.