Overview
Pivane is a self-hosted web workspace for AI coding agents, designed to be used from any screen. It runs the Pi coding agent on the user's own machine and allows access from any browser on desktop, tablet, or phone. The workspace supports subagents, scheduled tasks, MCP, long-term memory, rich file previews, and a built-in image, video, and speech lab, all working with the user's own models and keys.
Pivane uses Pi's native runtime and reads and writes Pi's own session files, providing streaming replies, tool calls, file diffs, shell commands, steering, and context compaction in a UI that is easy to read on any screen. Sessions are stored on the user's server, and when a thread is opened on another device, the same agent process serves it. Users can add Pivane to their home screen and get notified when a reply finishes.
The workspace allows running many agents from one place, including subagents, task threads that agents create for themselves, messages between threads, side chats, and cron-scheduled tasks. A live view shows what each one is doing. Users can use any provider Pi supports, including Anthropic, OpenAI, Google Gemini, GitHub Copilot, OpenRouter, DeepSeek, Qwen, Kimi, MiniMax, xAI, and Mistral, signing in with an API key or OAuth, or adding their own OpenAI-compatible endpoint. Credentials and history stay on the user's machine.
Pivane extends the same way Pi does: Pi packages, skills, prompt templates, and MCP servers can be installed from the browser, and a built-in extension assistant can find and set them up. Media generation and voice are built in, allowing connection to the user's own image, video, and speech services. An agent turns ideas into editable parameters, and nothing runs until the user confirms. Users can also dictate messages and have replies read aloud.
Pricing
Free
Detailed plans have not been confirmed in our catalog. Check the official website for current limits and billing terms.
Visit WebsitePrices and limits may change. Confirm the currency, billing period, seat minimum and usage caps on the official website.
Use Cases
- Teams use Pivane to run the Pi coding agent on their own machine and access it from any browser on desktop, tablet, or phone.
- Developers use Pivane to start a session on their desktop and check in from their phone, with sessions stored on their server.
- Users use Pivane to run many agents from one workspace, including subagents, task threads, messages between threads, side chats, and cron-scheduled tasks.
- Users use Pivane to connect their own image, video, and speech services for media generation and voice.
- Users use Pivane to dictate messages and have replies read aloud.
- Users use Pivane to install Pi packages, skills, prompt templates, and MCP servers from the browser.
- Users use Pivane to schedule tasks with cron-based or one-time schedules, supporting time zones, previews, budgets, and run history.
Features
Access coding sessions across devices
Manage multiple agents and subagents
Schedule recurring agent tasks
Connect MCP tools and extensions
Keep long term agent memory
Preview files and code changes
Create images video and speech
Dictate messages and hear replies
Reviews
No reviews yet. Be the first to share your experience!
FAQ
Is it a replacement for Pi CLI?
No, it is a companion. Both share the same Pi identity, so you can use whichever fits the moment.
Does it need a GPU?
No. Models run at your providers or on endpoints you configure.
Which languages are supported?
The interface is available in English and Simplified Chinese. Your conversations can be in any language your model supports.