Sidekick

Sidekick is a native macOS app for people who want AI answers grounded in their documents and research materials. Its built-in llama.cpp engine runs GGUF models such as Qwen3.5 without a separate model runner.

• Answers from your files: Group files, folders and websites into subject collections called experts, or attach files to a chat so the model can retrieve relevant material.
• Web research: Web search finds current information, while Deep Research reads webpages across multiple steps and combines findings into a report; both need internet access.
• Local agent actions: Function calling lets models perform calculations and take actions, including saving CSV files and creating drafts in your default email app.
• Canvas and visual output: Create, edit and preview code, websites and text, with edits to selected passages; responses can display LaTeX equations and charts, and you can drag charts into other apps.
• Conversation memory: The app remembers useful details and preferences you've shared for use in later chats.

Local model conversations work offline, with inference on your Mac. Optional remote models use your own API keys, with provider presets including OpenAI, Anthropic, Google AI Studio and DeepSeek. When you choose a remote provider, that service processes model requests.

Sidekick is open source under the MIT license and requires an Apple Silicon Mac with at least 8 GB of RAM. Text-to-image generation requires Apple Intelligence and macOS 15.2 or later.