Run AI on your own machine.

Chat with a model, ask questions about your own documents, search the web privately, make images, let it read web pages for you — all on your own computer. It never sends anything anywhere on its own.

Download for Windows macOS early Intel Mac early Linux early Free · Nothing to install first · Windows 10/11, macOS (Apple Silicon or Intel), Linux · New in 0.11.1 ↓
What your system will say about the download. The Windows installer is code-signed by Troy Grossi (right-click → Properties → Digital Signatures). Windows may still show "Windows protected your PC" for a while: SmartScreen trusts a new signature only after enough installs have gone well, and that takes weeks, not a setting. The Mac and Linux builds aren't signed, so macOS blocks the first open (System Settings → Privacy & Security → Open Anyway). Checksums are on the release page.

Come by the house.

Questions, bugs, ideas, or just to say what it ran on and how fast. The people building it are there.

Join the Discord

What it does

The chat: a question about an attached PDF, the steps it took, and the answer naming the page it read

Ask your documents

Attach a PDF and ask. It reads the page or section it needs, on this computer, and the answer says which page it read.

The Files page as a grid: PDFs show their first page, pictures their thumbnail, with folders on the left

Your files, in folders

Everything you gave it stays on this computer, in folders. A PDF shows its first page; a scanned page is read by the vision model, so it can be searched too.

A picture made with /image, with its prompt, seed, size and model beside it, and Edit this and Make another

Make and edit pictures

A picture keeps its recipe - prompt, seed, size and model - so you can edit it or make another. Optional: one click installs the picture engine and its models, about 13 GB to make pictures and another 29 GB to edit them.

Show the work: an /image command with every stage listed in words, from making room on the graphics card to saving the picture

Show the work

Every step as it happens, in words: which tool ran, what it read, each stage of a picture. Or switch to Answers only for just the replies.

The Tools page: /see, /image, /edit and /search, each with a switch, what it needs, and whether it asks before it acts

Every tool has a switch

What it needs, whether it is ready, and whether it asks before it acts. A tool that changes something outside the app asks you first, in the conversation.

The Test page: three test suites, each saying what it asks, what counts as a pass, and what it needs

See the test results

Every tool ships with what it scored - on which model, on what date - and the tests are included, so you can rerun them on your own hardware instead of taking our word for it.

Chat

A capable model running on your own machine. On first run it picks one that fits your hardware and tells you why; you can download others from the Models page.

Private web search

Search the web from the chat without a search engine watching. Switch it on and the app installs its own search engine, about 40 MB; nothing else to set up.

Let it browse

It can open a real web page and read it to answer you. It only clicks or types when you allow it, in the conversation.

Screenshots taken from the app on 2026-09-27, on a Windows 11 PC. Nothing staged.

What's new in 0.11.1

September 2026 · since 0.10.1

  • Notebooks: your notes are plain files on your own computer. Every note is a markdown file in a folder you can open, back up, or edit in any other program, and the copilot finds it there. Write notes beside your conversations, save any answer as a note in one click, or keep a web page you read as one. Make, rename and delete notebooks in the app; deleting one puts its notes in the trash, and Restore brings them back. Nothing is erased.
  • Files and Notebooks open over your conversation. The sidebar's Files and Notebooks open a window over the chat, with everything in it: folders, the trash, a file in its viewer, a note in its editor. Use in this chat puts a note in front of the copilot for your next message.
  • Keep typing while it answers. What you send while a reply is still arriving waits its turn under the message box, each with its own cancel, and goes as soon as the reply lands.
10 more changes
  • Stop really stops a picture. It used to say it had stopped, then hand you the picture anyway.
  • Your own files and notes come first. The copilot looks there before anything else, and goes to the web when the answer plainly lives there.
  • An answer is never a block of code it meant to run. If the copilot writes a step as its reply, you see a step that did not run, not raw JSON where the answer should be.
  • A file is kept and readable while search catches up. If making it searchable has to wait, the file says not searchable yet and becomes searchable by itself.
  • Pictures you make carry no prompt inside the file, and are named after what they show: cozy-reading-nook-blue-armchair, not create-an-image-of-a-cozy.
  • A web address without https:// is read, and a page that could only be read unencrypted says so.
  • The calculator is exact: 0.1 + 0.2 is 0.3.
  • Drop a file anywhere on the conversation to attach it, and the sidebar's Recent attaches a file in one click.
  • Commands sits beside the paperclip, New is a real button, and every menu is the app's own.
  • The model picker shows one model at a time in full: what it is good at and poor at, with its buttons always in view.

The full release notes

Before you install

The first run downloads a few gigabytes

The engine that runs models is 1.4 GB on Windows, and the model it chooses for you is between 1.4 and 13 GB depending on what your machine can hold — it names the size on screen before it starts. Measured end to end on a clean Windows 11 machine with no graphics card: installer to chat in 6m 04s, about 3 GB downloaded. Afterwards it starts in seconds.

It works without a graphics card, slowly

No card is required — it was tested on a machine with none. But the card is what makes it quick, and it decides how large a model you get: more video memory means a larger, more capable model; with none, it uses a small one on the processor. It tells you which it chose and why, and the Models page will install any of the others.

Windows is the tested one

The Windows installer has been run from scratch on a clean machine. The macOS and Linux downloads are the same application built for those systems; they exist and they are expected to work, and nobody has yet installed one on a clean machine and said so — that is what early means above. macOS comes in two builds — Apple Silicon (2020 and later) and Intel; Apple menu → About This Mac says which you have, and the wrong one refuses to open rather than running badly. Linux is an AppImage; a .deb is on the release page too. If one of them doesn't work for you, the release page is the place to say so.

What it needs

Graphics
None required; a card with 8 GB or more gets you a much larger model
Memory
8 GB
Disk
10 GB, considerably more for large models and image generation
System
Windows 10 or 11 · macOS, Apple Silicon or Intel · Linux (macOS and Linux are early)
Install first
Nothing. Everything optional is installed from inside the app, one click each