Show HN: Khoj – Chat offline with your second brain using Llama 2 (github.com)

Article summary

Khoj is a personal AI app that allows users to chat offline with their second brain using Llama 2, accessing information from the internet and local documents. It can be self-hosted or used as a cloud service, and has features such as semantic search, automation, and custom agents. The app integrates with various platforms, including Obsidian, Emacs, and WhatsApp. Khoj is open-source and aims to provide a private and customizable AI experience.

Main themes

• AI-powered personal assistants
• Lifelogging and searchable context
• Obsidian and Notational Velocity integration
• Large language models and customization
• Self-hosting and privacy
• Automation and productivity tools
• Traumatic brain injury support and accessibility

What commenters say

• Some users appreciate the Obsidian integration and suggest using Khoj with voice assistants or smart eyeglasses for a more seamless experience.
• Others discuss the potential benefits of lifelogging and searchable context for personal productivity and memory aid.
• A few users express frustration with the intentional delay in Obsidian's search function and appreciate the instant search update in Notational Velocity.
• There is a suggestion to use Khoj with a standardized API to serve various large language models, allowing for model swapping on the fly.
• Some users are interested in using Khoj for specific use cases such as traumatic brain injury support or automated research.
• The possibility of using Khoj with other models like llama-cpp-python is discussed, which could enable running larger models on custom hardware.
• The importance of privacy and self-hosting is highlighted, with some users appreciating Khoj's open-source nature and private data handling.