Native Rust harness
A native harness that travels with you
Run fast, token-saving native agents on your existing API keys and subscriptions — Claude, OpenAI, Gemini, Cursor, Copilot, DeepSeek, Kimi, and more. No margin, no lock-in, and your setup survives every model swap.
Bring your own key — and your own setup
Connect existing subscriptions and use them directly. Your MCP servers, skills, and the memories agents built about your codebase stay stable across sessions, model swaps, and agent upgrades.
Built in Rust, under 100MB
Tauri and a Rust backend instead of a bundled Chromium: a fraction of Electron's disk and RAM cost, for an app you leave running all day.
The ADE Manager
A built-in meta-agent that can spawn, monitor, and coordinate other sessions and drive the UI through a structured action system — and you can always just ask it for help.
Execution modes
Delegate fully, supervise at decision points, or pair in real time. Presence-aware routing behaves differently when you're watching, in a meeting, or away.
Resource-aware execution
Runs react to CPU, RAM, and your attention: heavy jobs pause on battery, and queued work continues overnight with scheduling and auto-started sessions.
On-device models
Through a research partnership with OpenBMB, local models power proactive housekeeping agents inside the app — groundwork for fully on-prem deployments.