Hi, I'm Teo a teenager that built GrgAI! It started because I was frustrated. Every time I wanted to use an AI assistant for serious work, I kept thinking: where is this data actually going? Who can read it? I couldn't find a good answer, so I decided to build something where the answer is simple: nowhere. It stays on your machine. I compiled CUDA kernels by hand, spent weeks debugging llama.cpp on a GTX 980, built a RAG system from scratch with 240+ knowledge files, and added a Privacy Mode specifically for people who legally cannot use cloud AI — lawyers, accountants. Is it perfect? No. It runs on a 3B parameter model and a GPU from 2014. But it works, it's private, and it's mine. If you're someone who's ever hesitated before pasting something into ChatGPT — this is for you. Would love any feedback, especially from people in regulated industries.

GrgAI

Your private AI.

Your private AI.