PI-Desktop: A local-first workspace for AI coding agents

Description and use

PI-Desktop provides a local-first desktop workspace that preserves AI coding session context, enables model switching, manages permissions, and supports multi‑agent workflows.

Problem

AI coding sessions lose context between restarts, making it hard to track progress, manage permissions, and coordinate multiple agents.

Outcome

PI-Desktop enables developers to maintain persistent AI coding sessions, switch models freely, manage permissions, review diffs, and coordinate multi-agent workflows within a single local-first desktop application.

Workflow

• Launch PI-Desktop on macOS/Windows/Linux
• Open or create a project to store context
• Start an AI coding session; the workspace retains conversation history
• Switch between model providers or models without rebuilding workflows
• Use Plan and Goal modes to outline tasks and track progress
• Set permission controls to approve destructive actions
• Review diffs before accepting changes
• Delegate tasks to subagents via multi-agent workflows
• Extend functionality with plugins and MCP integrations

Evidence

PI-Desktop is built with Electron, a Rust host core, and the pi agent harness.

Source

Persistent projects and sessions — project context and conversation history stay available across restarts.

Source

Model freedom — switch providers and models without rebuilding the whole workflow around one vendor.

Source