Skip to main content
PromptQuorum
Home/Power Local LLM/Sidekick Review 2026: A Free, Native macOS Local AI Chat App
Easiest Desktop Apps

Sidekick Review 2026: A Free, Native macOS Local AI Chat App

Β·10 min readΒ·By Hans Kuepper Β· Founder of PromptQuorum, multi-model AI dispatch tool Β· PromptQuorum
Read in:
πŸ‡ΊπŸ‡ΈenπŸ‡©πŸ‡ͺdeπŸ‡«πŸ‡·frπŸ‡―πŸ‡΅jaπŸ‡¨πŸ‡³zhπŸ‡ͺπŸ‡ΈesπŸ‡§πŸ‡·ptπŸ‡ΈπŸ‡¦arπŸ‡°πŸ‡·ko

Sidekick is a free, open-source, native macOS app that lets you chat with a local LLM that can respond using information from your own files, folders, and websites, without installing any other software. Created by developer John Bean (johnbean393) and released under the MIT license, Sidekick runs entirely on Apple Silicon Macs through a built-in llama.cpp engine, and can also connect to cloud providers such as OpenAI, Anthropic, Google AI Studio, DeepSeek, and Groq through their APIs.

Sidekick (github.com/johnbean393/Sidekick) is a free, open-source, native macOS application that chats with a local LLM and can pull in information from your own files, folders, and websites, with no other software required. It is built on llama.cpp for local inference and is designed specifically for Apple Silicon Macs, with over 3,300 GitHub stars. This review covers what Sidekick actually does, how to install it, and who it fits.

Key Takeaways

  • Sidekick (github.com/johnbean393/Sidekick) is a free, open-source, native macOS app β€” not a cross-platform Electron app
  • Created by developer John Bean (GitHub handle johnbean393)
  • MIT licensed, confirmed via the repository's LICENSE file
  • Requires an Apple Silicon (M-series) Mac β€” this review found no Intel Mac, Windows, or Linux build
  • Runs local models through a built-in llama.cpp engine, with retrieval-augmented generation over your own files, folders, and websites via customizable "Experts"
  • Can optionally connect to cloud providers β€” OpenAI, Anthropic, Google AI Studio, DeepSeek, Groq β€” through OpenAI-compatible API connections
  • Over 3,300 GitHub stars as of this review, per GitHub API

πŸ“ In One Sentence

Sidekick is a free, open-source (MIT), native macOS app for Apple Silicon Macs that chats with a local LLM through a built-in llama.cpp engine and can answer questions using your own files, folders, and websites, with over 3,300 GitHub stars.

πŸ’¬ In Plain Terms

Sidekick is a Mac app you download, install, and run without a terminal or Python setup. It downloads an AI model to your own computer and lets you point it at specific folders, files, or web pages so it can answer questions about them β€” no cloud account or subscription needed, though you can optionally add a cloud API key later.

πŸ“ŒNote: This review is the deep-dive companion to Sidekick's entry in the Local LLM Software Directory β€” see that page for how Sidekick compares at a glance to dozens of other local AI tools. It is based on Sidekick's own GitHub repository, README, and setup guide, not hands-on benchmarking by PromptQuorum.

What Is Sidekick?

Sidekick is a native macOS chat application that runs a local LLM on your own Mac and can pull relevant information from your own files, folders, and websites into its answers, without needing any other software installed. Its own GitHub description states it lets users "chat with a local LLM that can respond with information from files, folders and websites on your Mac."

  • Product type: a native macOS desktop application, not a browser app, CLI tool, or IDE extension
  • Creator: developer John Bean, GitHub handle johnbean393
  • Repository: github.com/johnbean393/Sidekick
  • License: MIT, confirmed via the repository's own LICENSE file
  • Platform: Apple Silicon (M-series) Macs only, per the project's own setup guide β€” this review found no Intel Mac, Windows, or Linux build
  • Scale: over 3,300 GitHub stars as of this review, per GitHub API

What Does Sidekick Actually Do?

Sidekick runs a local LLM on your Mac and layers retrieval-augmented generation, tool use, and a few other features on top of it, all built into one native app.

  • Local-first chat β€” runs GGUF models locally through a built-in llama.cpp engine, so conversations can happen fully offline once a model is downloaded
  • "Experts" (retrieval-augmented generation) β€” point Sidekick at specific files, folders, or websites, and it retrieves relevant passages into its answers, entirely on-device
  • Cloud provider connections β€” optionally add your own API key for OpenAI, Anthropic, Google AI Studio, DeepSeek, Groq, or other OpenAI-compatible endpoints, to use cloud models alongside local ones
  • Function calling β€” Sidekick's own documentation describes sequential tool execution, letting the model call functions to perform calculations or actions as part of a response
  • Deep Research β€” a multi-step research agent that Sidekick's own project description says can synthesize information across 50–80 webpages for a single query
  • Memory β€” retains information about the user between separate conversations, per the project's own feature list
  • Canvas editor β€” a built-in editor for creating and selectively editing websites and code, rather than only receiving code in chat form
  • Image generation β€” automatic detection and image generation via CoreML on macOS 15.2 and later, per the project's documentation

Usage Examples: Three Ways to Use Sidekick

These are concrete workflows built from Sidekick's documented features above β€” not hypothetical use cases.

Platform, Pricing, and Licensing

Platform

What Sidekick states:
A native macOS app for Apple Silicon (M-series) Macs only, per the project's own setup guide β€” no Intel Mac, Windows, or Linux build.

Cost

What Sidekick states:
Free and open source. Cloud provider connections are optional β€” you pay that provider directly if you add an API key; Sidekick itself has no paid tier.

Licensing

What Sidekick states:
MIT license, confirmed via the repository's own LICENSE file.

Requirements

What Sidekick states:
Apple Silicon Mac with at least 8 GB RAM; image generation additionally requires macOS 15.2 or later, per the project's documentation.

Verify current requirements and licensing directly on github.com/johnbean393/Sidekick before relying on this table for a compliance or purchasing decision.

Who Should Use Sidekick?

Sidekick fits Mac owners on Apple Silicon who want a native, no-terminal way to chat with a local model over their own files without installing any other software.

What Sidekick Is Not Good For

Sidekick is not a fit if you are not on Apple Silicon, or if you need a mobile app or enterprise-grade support.

  • Not for Intel Macs, Windows, or Linux β€” this review found no build for any platform other than Apple Silicon macOS
  • Not a mobile app β€” there is no iOS or Android version documented in the repository
  • Not backed by a company this review could verify β€” treat Sidekick as an independently maintained, community-supported project rather than a venture-backed product with a guaranteed long-term roadmap
  • Not a fixed-cost tool if you add cloud providers β€” connecting OpenAI, Anthropic, or another API key means your costs scale with that provider's own pricing, separate from Sidekick itself
  • Not built for enterprise deployment β€” this review found no team-account, centralized-admin, or SLA offering

Common Mistakes When Evaluating Sidekick

Most confusion about Sidekick comes from conflating it with unrelated same-named projects, or assuming it supports hardware it does not.

Competitors and Alternatives

Within native, local-first Mac chat apps, Sidekick is most directly comparable to Enchanted and BoltAI, plus Msty for a cross-platform alternative and Khoj for a document-focused competitor with a similar retrieval-augmented approach.

BoltAI

Best known for:
Native macOS AI chat app with system-wide shortcuts and multiple provider support
Link:
BoltAI Review
Articles about BoltAI (1)

Also mentioned in:

Msty

Best known for:
Cross-platform (macOS/Windows/Linux) local and cloud chat client with built-in RAG
Link:
Msty Review
Articles about Msty (5)

Also mentioned in:

+3 more not shown

This is not an exhaustive list of local chat apps β€” see the Local LLM Software Directory for the full, regularly updated catalog, including Sidekick's own directory entry.

Frequently Asked Questions

What is Sidekick?

Sidekick (github.com/johnbean393/Sidekick) is a free, open-source (MIT), native macOS app that chats with a local LLM and can answer questions using your own files, folders, and websites.

Is Sidekick free?

Yes, Sidekick itself is free and open source under the MIT license. Connecting an optional cloud provider (OpenAI, Anthropic, etc.) costs whatever that provider charges you directly; Sidekick does not add a fee on top.

How do I install Sidekick?

Per the project's own documentation, run brew install --cask arcadi4/tap/sidekick via Homebrew, or use the direct download linked from Sidekick's GitHub repository. Both require an Apple Silicon Mac.

Does Sidekick work on Intel Macs, Windows, or Linux?

No. This review found no Intel Mac, Windows, or Linux build β€” Sidekick is built specifically for Apple Silicon (M-series) Macs.

Does Sidekick require an internet connection?

No, not for core chat and file/folder retrieval β€” once a model is downloaded, Sidekick runs entirely on-device through its built-in llama.cpp engine. An internet connection is only needed if you add an optional cloud provider API key.

What local models does Sidekick support?

Sidekick runs GGUF models through its built-in llama.cpp engine, including modern model families such as Qwen, per the project's own documentation. Check the GitHub repository for the current list of tested models.

Who created Sidekick?

Developer John Bean (GitHub handle johnbean393) created and maintains Sidekick as an independent open-source project.

Can Sidekick answer questions about my own files?

Yes, through its "Experts" feature β€” retrieval-augmented generation over files, folders, or websites you point it at, processed entirely on-device.

What is Sidekick's minimum hardware requirement?

An Apple Silicon (M-series) Mac with at least 8 GB of RAM, per the project's own setup guide. Which specific local models run comfortably depends on that model's own size relative to your Mac's memory.

Has PromptQuorum independently tested Sidekick's feature claims?

This review is based on Sidekick's own GitHub repository, README, and setup guide, rather than hands-on benchmarking by PromptQuorum.

Sources

← Back to Power Local LLM