Do I need an internet connection to use Billy?

No. Once Ollama and your chosen model are downloaded, Billy works entirely offline.

Which models does Billy support?

Billy works with any model available via Ollama, including qwen2.5-coder, codellama, deepseek-coder, llama3, and mistral.

How is Billy different from GitHub Copilot?

Billy runs locally via Ollama with no subscription fee and no cloud dependency. Your code stays on your machine, and Billy can run shell commands with your approval.

Does Billy work on Windows?

Yes. Billy works on Windows via Scoop or the downloadable Windows binary from GitHub Releases.

How does Billy make money if the app is open?

Billy is supported through optional setup help, sponsorship, and future convenience bundles. The core local app remains usable without a purchase.

Billy — Local AI Coding Assistant for the Terminal

Billy is a local AI coding assistant for the terminal and a GitHub Copilot CLI alternative that runs fully offline via Ollama. No subscription, no API keys, no data leaving your machine. The core app ships open and is supported through optional setup help, sponsorship, and future convenience bundles.

Why Billy?

Fully Private — All AI runs on your machine via Ollama. Zero telemetry, zero network calls during inference. Works offline.
No Subscription — Billy ships open-core by default. Support the project through optional setup help or sponsorship instead of a forced paywall.
Agentic Mode — Billy proposes shell commands and runs them with your approval. Reads output, self-debugs, iterates until done.
Memory System — Say "remember that I'm using Go 1.24" and Billy saves it. Injected into every conversation automatically.
Session History — Full conversation history. Resume any past session with /history. Context compaction with /compact.
Model Switching — Switch any Ollama model on the fly. Pull new models without leaving the app.

Install Billy

Desktop (macOS / Linux): curl -fsSL https://raw.githubusercontent.com/jd4rider/billy-wails/main/scripts/install.sh | bash

CLI (macOS / Linux): curl -fsSL https://raw.githubusercontent.com/jd4rider/billy-app/main/scripts/install.sh | bash

Homebrew: brew tap jd4rider/billy && brew install billy

Scoop (Windows): scoop bucket add billy https://github.com/jd4rider/scoop-billy && scoop install billy

Desktop downloads: macOS .pkg / .tar.gz, Windows .zip, plus CLI release archives on GitHub Releases.

Support Model

Free Core — Full local CLI, agent mode, memory, history, and custom backends with DIY setup.
Setup Help — Optional guided install and troubleshooting if you want someone to wire Billy up for you.
Support the Project — GitHub Sponsors and Buy Me a Coffee help keep the apps ad-free and local-first.
Future Convenience Bundles — Optional scripts, workflow packs, or specialty integrations may come later.

The core app stays open. Paid paths are for convenience and sustainability, not artificial lockouts.

Billy vs GitHub Copilot and Cursor

Billy is built for developers who want a local AI coding assistant in the terminal instead of a monthly subscription. It runs offline with Ollama, supports agentic shell workflows, and keeps your code on your own machine.

Frequently Asked Questions

Do I need an internet connection to use Billy?: No. Once Ollama and your chosen model are downloaded, Billy works entirely offline.
Which models does Billy support?: Any model available via Ollama — including qwen2.5-coder, codellama, deepseek-coder, llama3, mistral, and hundreds more.
How is Billy different from GitHub Copilot?: Billy runs 100% locally via Ollama with no subscription fee. Your code never leaves your machine. It also has an agentic mode that can run shell commands with your approval.
Does Billy work on Windows?: Yes — install via Scoop for the CLI or download the Billy Desktop Windows bundle from GitHub Releases.
How does Billy make money if the app is open?: Billy is supported through setup help, donations, sponsorship, and future convenience bundles. The goal is to keep the core app usable without ads or a paywall.