One of acquaintances asked me a simple question - how do we connect? I didn t want to provide my mobile number and zoom free has limitations. Then thought why do we need these for a simple 1:1.So created a meeting tool using webrtc - this doesn t need any account setup or password as it is P2P. Backend doesn t store anything. Web recording stays local. Can share the screen. Works on mobile and web. Can schedule and add it to your calendar as well.Once I got that working, I was previously doing a spike on a dictate/note taking tool - in the same spirit - no accounts/no password - everything local using web gpu and small LLM. To my surprise small LLM works well for this usecase.Now that I got that working, I then integrated the AI note taking tool into the meeting tool with an iframe and now I have a simple, no frills meeting tool that doesn t feel too basic. ____________After that my son wanted to create personalized bedtime stories - yes he wants to be the hero. But gemini, claude, chatgpt all don t create really long form bed time stories - the voice option stops after 1 min or so for you to keep saying continue and even then the length of the story is not long.And I didn t want the personal info going into any cloud memory.So with the same concept of no accounts/no passwords, all local using webgpu - created story maker. Had to work for sometime to get the chunking strategy work well with small LLM but now it creates long story and then used supertonic 3 for narration. For all small LLMs I am using ONNX.Along the way also created simple tools - web reader, document reader. Also one more fun chat agent - AI spirits - it is just the local LLM with personality - a mini/micro Characters.ai if you will. All these tools are at voice.cogumi.comOne more - actually two - image generation from prompt and art from doodle tools, with local LLM. But this one is a bit slow.Try it and let me know your comments here or directly at hello@cogumi.comI am planning to just create such pay it forward tools. Think, like me a lot of people want to cut on subscriptions. Many get stymied managing the local env to get the local LLM going. So here is my attempt at it to solve.