Hey Product Hunt 👋 I built KathGPT because I wanted a serious AI workspace on my desktop — without handing my conversations, workflows, and API keys to yet another cloud service. The problem: Most AI apps make you choose between convenience and control. You either use a hosted chat product (your data lives on their servers) or stitch together browser tabs, API keys, and local scripts. I wanted something that felt as polished as a modern chat app, but where everything stays on my machine — chats in SQLite, keys in the OS keychain, API bound to Local only. What KathGPT is: A native desktop app (macOS, Windows, Linux) for multi-model chat, research with citations, image generation, PDF translation, workflows, and a prompt library. Bring your own key — OpenRouter, OpenAI, Anthropic, Gemini, or Perplexity. No KathGPT account, no subscription, no telemetry. How it evolved: KathGPT started as a web app with a Node backend. For v0.1 “Local Edition” I rewrote the entire API in Rust (Axum + SQLite) and wrapped it in Tauri v2 — one native process instead of Electron + a separate server. The goal was simple: faster streaming, smaller installers, and data that never leaves your disk unless you send a prompt to an LLM provider. It’s open source (MIT) — inspect it, fork it, run it yourself. Try it: Download from our site or GitHub Releases. Add an API key in Settings and you’re chatting in under a minute. I’d love your feedback — especially on privacy, performance, and which tools to build next. What would make this your daily AI app?

KathaGPT

OpenSource- Private AI workspace, built for your desktop

OpenSource- Private AI workspace, built for your desktop