VivianneVivianne

The agent harness for code, math, and science.

Vivianne turns any capable language model into an AI agent that works on your computer: it runs code in a real terminal, delegates to subagents, and remembers your projects, with the model of your choice or one running locally.

A master that runs the team. Hand off the project and it staffs specialist agents that do the work or run swarms, while you make the decisions.

Vivianne's Agents panel: a Project Versa Lead master agent with a team of four employee agents (Build & Tooling, Quality & Evaluation, Renderer, and Runtime Integration engineers), each scoped to its own files and tools and reporting to the master
Coming soon

Multiplayer. Pair two Vivianne instances and build together: see what the other agent did, and split the work in one shared chat.

  • Pair with a short code. Direct, encrypted peer-to-peer, no account or server in the middle.
  • Follow your pair's session as it happens, turn by turn.
  • Chat with your pair, send code snippets, and divide up the work.
Built for STEM work

One AI agent for coding, mathematics, science, and technology.

Same harness, same tools, same memory, whatever the problem is. Coding is where it is strongest today; math and science workflows are being built on top of it now.

Live

Coding

AI coding agent

Reads your whole codebase, edits across files, runs builds and tests in a real terminal, and shows every diff in a Review panel before you keep it.

In progress

Mathematics

AI agent for math

Works a problem the way you would: sets it up, writes and runs code to check each step numerically, and plots the result instead of guessing at it.

In progress

Science

AI for scientific research

Literature review, data analysis, and reproducible experiments, orchestrating the scientific tools you already use rather than replacing them.

Live

Technology

AI agent for developers

Terminal, browser, MCP servers and your file system as first-class tools, with an approval step on anything that touches your machine.

Programmatic tool calling

Instead of one round-trip per tool, it writes a short script that calls its own tools inline. Twenty file reads collapse into one step, and every call still goes through your approval.

Extend it with playbooks

Drop-in add-ons that give the app new tools and their own panels, for the workflows you repeat.

Hands you a real file

Ask for a report, spreadsheet, or deck and it writes an actual PDF, Word, Excel, or PowerPoint file.

Any model, cloud or local

Bring the model. Or run it on your own machine.

Configure a key once in Settings and switch models mid-session. Or skip the key entirely: point Vivianne at a GGUF file on disk and it runs the model itself, fully offline, with nothing sent anywhere.

  • Anthropic
  • OpenAI
  • Google
  • Groq
  • xAI
  • DeepSeek
  • Mistral
  • Together AI
  • Fireworks
  • OpenRouter
  • Cerebras
  • NVIDIA NIM
  • Hugging Face
  • Cloudflare Workers AI
  • Local GGUF
  • Any OpenAI-compatible API
Vivianne's Settings → Providers panel, showing the full list of connected model providers

Put a real agent on your desktop.

Download for WindowsWindows today · macOS and Linux on the way