Sidekick is a free, open-source native macOS app that lets you chat with a large language model running entirely on your own Mac. Powered by llama.cpp, it can answer questions using your local files, folders and websites as context — with no server, no account and nothing leaving your machine, so it works fully offline and privately.
Sidekick is an open-source project on GitHub with about 3,300 stars, written in Swift and actively developed in 2026. Unlike Electron chat clients, it is a true native SwiftUI app that runs open-weight models locally through llama.cpp — including DeepSeek R1, Llama and Qwen families. Point it at folders or documents and it uses retrieval over your content so answers can cite and use your own material; it can also pull in website content for context. Inference features such as grouped-query attention and efficient KV caching keep local generation responsive on Apple Silicon. Because everything runs on-device, Sidekick suits anyone who wants an AI assistant over personal or work documents without sending data to the cloud.
Sidekick is completely free and open source under the MIT license. Download it or build it from source — there are no subscriptions or paid tiers.
Sidekick is for Mac users who want a private, offline AI assistant that can chat about their own documents — researchers, writers and professionals who cannot or do not want to send files to cloud AI services.