Day one on Product Hunt with no audience + a local model question
I launched Awnsy about 15 hours ago. It's a menu bar translator for macOS that runs the model on your Mac: select text, press ⌘C twice, translation shows up right there. No newsletter, no followers, no team, first Mac app. Most launch posts I read here come from people with all three, so here is the other kind.
The numbers
Currently 12 upvotes, 3 comments, #17 for the day.
Downloads and promo code redemptions: I don't know yet. Awnsy has no analytics SDK at all. That's what lets me keep the "Data Not Collected" privacy label on the App Store, and for an app whose whole point is that your text stays on your Mac, I'd rather have that label than a dashboard. App Store Connect's own numbers haven't updated yet. If anyone's curious, I'll post them here once they do.
One thing I didn't expect: the launch filled my inbox. Offers started coming in by email and on LinkedIn, and the PH page made it easy to get listed elsewhere. Awnsy is now on MacAIApps, LaunchBuff, LaunchBuck, Stork.ai and a few others, and one of those added it without me asking.
What I learned
The first review came from another founder, a long-time DeepL user. What sold him wasn't model quality. It was that the translation shows up right where you're already reading, with no switching to a browser tab, and that the local model needs no account or API key, so he didn't have to think twice before pasting something sensitive. The double ⌘C got a mention too: no separate hotkey to remember. He also asked for a progress indicator in the screen area mode, and that's what I'm building this week.
So the lesson from day one: people pick a translator for how it fits into what they're doing, and for trust. That's the part I'll keep investing in.
If you try Awnsy, tell me what you think. I read everything, and the very first review already shaped this week's work.
And one question for anyone here who runs models locally: have you tried any of the translation models on Hugging Face? Awnsy's default is Hy-MT2 1.8B, about a second per sentence on an M1 MacBook Air. The 4B models I tested scored about the same but ran 2 to 5 times slower. I'm picking the next model to add, and I'm looking for something fast that runs locally on an M1 and up. Which one would you try?
Text translation is free. PRODUCTHUNT gives 3 months of Pro until Oct 19, for new subscribers.
Replies