Launching today
jotty.pro
Private AI workspace powered by in-browser AI models
1 follower
Private AI workspace powered by in-browser AI models
1 follower
Your train of thought has a place to stop. Real-time assistance pays attention during interviews, meetings, and speaking practice. Your adaptive homepage turns what you've captured into something valuable you can use. With local LLMs running in your browser, you can work on personal notes without sending them elsewhere.
Interactive
Launch Team
Framer AI AgentsDesign and publish professional sites with AI
Promoted
Hey Product Hunt 👋
I kept wondering why speech-to-text (meeting notes, dictation apps, name it) tools need to be so heavy. The browser already has access to your microphone, can capture audio from your tab/window/device, and can store your notes. Why install and maintain another app for that?
AI processing used to mean sending something to the cloud. Now models can run directly in a browser tab. That's the core of what I'm building.
You can record, transcribe, and manage your notes right in your browser. Your notes stay encrypted on your device, so you’re in charge of backups. There’s no need to install anything, use a meeting bot, or send your data to third parties unless you want to.
Everything else grew around that.
Real-time assistance helps during interviews, meetings, and speaking practice. It follows the conversation, your template if any, and helps you stay on track.
The homepage adapts to you, making your captured notes worth coming back to. It highlights useful ideas, questions, and follow-ups, and you decide what shows up. I don’t want just another spot to collect notes and forget about them.
[My son helped make that awesome 6–7 video in the carousel. We’re both really proud of it 😄]
There are still some limits. Not every laptop can easily run both speech-to-text and a text generation model at the same time. You can connect to cloud providers using your own API keys, and the live assistant uses that method for now. I’m working toward full local processing, with practical options available as I go.
Having said all that, I'm really hoping to find more people who want to push this forward:
Help make in-browser LLMs more common – I’d love to connect with anyone building, experimenting, or helping others get started.
Test on-device text generation – I’d love to see more people trying out what works on different devices. I reached out to Liquid AI about running LFM2.5 with WebGPU, but haven’t been able to connect yet. The model still seems promising.
Build a community around what works. I want users to test and share their own templates for different use cases and models, and help grow the platform’s stats. A website full of prompts isn’t enough for me. Templates backed by the community, with proof of how they perform across models, really stand out.
If you’re trying out any of these ideas, I’d love to swap notes.
Thanks for taking a look 🙏