HyperContext

A minimalist chatting tool for chatting and reading on webpages

Learn more on hypercontext.org Note: This tool is not a product and will not be improved upon significantly. It's more of a demo of some interesting interfaces, explored in this blog: https://www.tomjo.me/hypercontext Details: - Your preference of GPT-3.5 or GPT-4 - OpenAI API key required - Completely free besides direct OpenAI chats and embeddings - Totally local storage and no sign-in required Features: - Have a selection of text explained. - Ask a question manually, in the context of a selection. - Converse with a chatbot that outputs citations. When a citation is clicked, its snippet automatically scrolls into view. - Generate key details to get a summary of the webpage. - Select text and attach a persistent annotation to it. That annotation will stay there the next time you visit the webpage. Getting Started: 1. Setup an OpenAI API key if you haven’t already. This will require putting in billing info on OpenAI's website. 2. Download the extension 3. Open the sidebar (Ctrl + E or click extension icon) 4. Settings Icon > Set OpenAI Key > Input your key > Press Save 5. Select any text and try a chat action or open the chat sidebar (Ctrl + E) Important Notes: - Cost: GPT-4 can get pricey. You can see the approximate cost of each chat by clicking the dollar sign icon in the chat header. Embeddings on the other hand are very cheap. - Document Length: Can handle documents up to 30 pages long. Uses RAG under the hood. - Sharing: Chats and annotations cannot be shared right now. To learn more, visit hypercontext.org