Sidekick

Sidekick is a native macOS app for people who want AI chat grounded in their own documents and research materials. Its built-in llama.cpp engine runs GGUF models such as Qwen3.5 without separate model software. Local conversations work offline. The app is open source under the MIT license and requires an Apple Silicon Mac with at least 8 GB of RAM.

You can group files, folders and websites into subject-specific collections called experts. Sidekick retrieves relevant material from the active experts to help answer questions, and it also accepts files dragged into a conversation. Memory carries useful details and preferences between chats.

Web search gives it access to current information, while Deep Research reads webpages and combines findings into a report across multiple research steps. These web features need internet access. You can also use remote models through your own API keys, with provider presets including OpenAI, Anthropic, Google AI Studio and DeepSeek. With a remote provider selected, that service processes model requests rather than the built-in local engine.

Function calling lets local AI agents perform calculations and actions, including saving CSV files or drafting email in your default mail app. Canvas supports creating, editing and previewing code, websites and text, with targeted edits to selected passages. Chat responses can display LaTeX equations and charts that you can drag into other apps. Text-to-image generation requires Apple Intelligence and macOS 15.2 or later.