Chatforge launches open-source local AI chat that merges conversations and builds persistent memory — no data leaves your computer

A developer released Chatforge, a free open-source chat interface that lets you drag two separate AI conversations together to combine their context, then ask questions that span facts from both chats. It runs entirely on your own GPU using Ollama (a local AI tool) and stores everything in a local database — nothing syncs to the cloud.

Unlike typical AI chat apps that treat each conversation as isolated, Chatforge automatically extracts and remembers facts about you after every exchange, then injects that memory into all future chats without eating up the AI's limited context window. A future feature (Phase 4) will compress merged conversations into the AI model's weights themselves, permanently baking in knowledge so you never lose important context.

This matters for anyone who wants an AI assistant that learns about them over time without sending data to Anthropic, OpenAI, or Google. Researchers, privacy-conscious professionals, and people managing sensitive work can now build a genuinely personal AI without trade-offs — the code is on GitHub (MIT license), requires Python 3.9+, Node.js 18+, and an NVIDIA GPU (or CPU, but slower).