Gergy is a desktop AI assistant that turns typed requests, dropped files, pasted notes, and optional voice commands into real project execution with Claude, Codex, Copilot, or Gemini.
Chat-first workflow · Optional voice and wake word · macOS & Windows
Pick your installed CLI, choose chat or voice, and keep everything grounded in real project folders on your machine.
Switch between Claude, Codex, Copilot, and Gemini without changing your workflow. Gergy's adapter layer normalizes all four into a single consistent interface.
Start with a dedicated chat window, drag files straight into the request, paste screenshots, or drop long notes that Gergy stages directly into the project workspace.
Prefer hands-free control? Enable voice and wake word in onboarding or settings. Prefer chat only? Leave them off and use the same execution engine without the avatar flow.
First launch walks you through CLI detection, voice preferences, avatar choice, and wake-word settings. Every session stays in SQLite-backed history so you can reopen it later.
Gergy normalizes Claude, Codex, Copilot, and Gemini into a single consistent execution layer. Switch anytime without changing your workflow.
Deep reasoning & tool use
The default adapter. Claude's extended context and native tool calling make it ideal for complex, multi-step projects that read and write files.
Code-native intelligence
High-reasoning execution in a sandboxed workspace. Codex's native code understanding and execution make it uniquely suited for agentic coding tasks.
GitHub subscription included
Use your existing GitHub Copilot subscription. No additional API key required — streaming JSON responses with minimal setup.
Multimodal frontier model
Google's latest multimodal model with massive context and structured JSON streaming. Outstanding for research, analysis, and vision tasks.
Gergy works with any supported assistant already installed on your machine. Pick one and follow the steps — it takes under two minutes.
Deep reasoning & tool use
npm install -g @anthropic-ai/claude-code
claude
Code-native intelligence
npm install -g @openai/codex
codex login
codex --version
GitHub subscription included
npm install -g @github/copilot
copilot auth login
copilot --version
Multimodal frontier model
npm install -g @google/gemini-cli
gemini
Installed CLIs are auto-detected, the welcome flow sets your defaults, and every request can begin in chat without touching a terminal.
Pick any supported assistant: claude, codex, copilot, or gemini. Gergy auto-detects what's available on your PATH.
On first run, Gergy detects installed CLIs, defaults to the first one it finds, and lets you decide whether to enable voice or stay chat-first.
Open a dedicated chat window, describe what you want, and optionally drag in files, paste images, or drop long notes that should live in the project folder.
Gergy binds the request to a project, starts the selected CLI, streams execution activity live, and opens a preview or folder view whenever you need it.
Download the desktop app, install one supported CLI, and choose chat-only or voice-enabled control during first-run setup.
Open the .dmg and drag Gergy.app to Applications. If macOS says the app is damaged or unverified, open Terminal and run:
xattr -cr /Applications/Gergy.app
Then double-click the app to launch normally.
Edge or SmartScreen may warn the file is unrecognized. Click Keep in the download bar, then Keep anyway in the SmartScreen prompt to proceed with installation.
Requires at least one AI CLI installed on your system. Use the install guide above to set up Claude, Codex, Copilot, or Gemini.