Johnbean393 / sidekick: a native macos app that allows users to chat with a local LLM that can respond to files, your Mac files without setting up any software. Llama.cpp operated.

Chat a Local LLM that can answer with information from your files, folders and website to your Mac without installing any other software. All conversations happen offline, and your data is saved locally.

Let’s say you’ve collected evidence for a paper in history about interactions between Aztecs and Spanish troops, and you are looking for the Aztecs who have taken Spanish weapons.

Here, can you ask Sidekick, “Does the Aztec use the acquired Spanish weapons?”, And it responds to direct quotes on page numbers and a short analysis.

To determine Sidekick response, simply click on the references shown by Sidekick’s low response, and the academic paper mentioned in the sidekick open immediately to your viewer.

Sidekick accesses files, folders, and websites from your experts, which can be individually configured with resources in relation to specific areas of interest. Activating an expert allows sidekick to access relevant information.

Since Sidekick used rag (Retival aughented generation), you can place theoretically to place unlimited resources with each expert in relation to your request. It is in the sharp difference in most services, including OpenII’s ChatGPTwhich can only ingest ~ 45 pages in the text.

For example, a student can make experts English Literature,, Mathematics,, Geography,, Computer Science and Physics. In the image below, he activates the expert Computer Scienceallowing the sidekick to respond to information from that expert.

Also can also provide users to access sidekick files by simply dragging the field.

Sidekick can still answer the latest information using Web Searchfacilitating research.

Sidekick supports various models of reasoning, including qwq-32b and Dereseek by Alibaba-32B and Dereseeek-R1.

Sidekick uses a code translator to enhance mathematical and logical models.

Since small models are better at writing the code than making maths, writing it in the code, implement it, and present the results that increase reliably with answers.

Sidekick can create images from the text, allowing you to create visual aids for your work.

No buttons, no flick switches, no Image Generation mode. Instead, a model shop at Coreml automatically indicates The image generation prompts, and gives an image as needed.

Image generation available on MacOS 15.2 or above, and requires the apple intelligence.

Press Command + Control + I to access Sidecick inline to write Sidekick Writing. For example, use the Answer Question Command to do your homework without leaving Microsoft Word!

The mark of the translated beautiful on the sidekick.

Sidekick offers native latex to give for mathematical equations.

Delayed roads automatically performed for tables when appropriate, with different charts available, including bar charts, line charts.

Charts can be dragged and fall into third-party apps.

The code is beautifully provided by syntax highlighting, and can be exported or copied by clicking a button.

use tool In the Sidekick of Supercharge your workflow.

Use the detector to evaluate the percentage of AI text, and use suggestions suggestions to rewrite AI content.

The diagrammer allowed you to advance in the intricate relationship relationships all from one quick. Take advantage of integrated preview and editor for easy editing.

Instead of making a powerpoint, just write a quick one. Use AI in Craft 10 Minutes Presentations in 5 minutes.

Export of common formats such as PDF and PowerPoint.

Sidekick used llama.cpp As the backend arrives, optimized by the lightning delivery of easy generation accelerated by Apple Silicon. Sidekick also supports dispoding decoding, which can increase performance up to 51%.

Optional, offload Genese to facilitate generation while raised battery life on your MacBook.

Requirements

• A Mac with Apple Silicon
• RAM ≥ 8 GB

send

• Download the disk image from releasedand open it.

Sidekick’s primary purpose is to make open, local, private models access to many people, and allow a local model to gain context with selected files, folders and websites.

Sidekick is a native application of LLM for macos who are fully running locally. Download it and ask your llm a question that doesn’t do any insertion. Give access to LLM in your folders, files and website with only 1 click, allowing them to respond to context.

• No config. Available to people who have never heard of models, prompting, or llms.
• Performance and simplicity of developer experience or parts. Not word, poor electron.
• Local first. Core functionality should not require an internet connection.
• No tracking of conversation. Talk about anything you want with sidekick, like notes.
• Open source. What’s the point of running local AI if you can’t audit it really runs locally?
• CONTACT TO START. Your files, folders and content are known.

Contributions are very acceptable. Let’s make the sidekick simple and powerful.

Talk to the owner of this repository to Johnbean393@gmail.comor filing an issue.

This project cannot be done without the hard work of:

• Psugihara and Antentirors who built Freechatwhere this project caught heavy inspiration from
• Georgi Gerganov for Flame.cpp
• Alibaba for training qwen 2.5
• Meta for training Llama 3.1
• Google for training in Gemma 2