Thursday
An open-source voice assistant that runs on your own computer and your own OpenAI key: you talk to it, and a background team of AI bots handles slower work in a real browser, a shell and your files while the voice call keeps talking.
Thursday runs a live voice call using a real-time voice model, with a separate model handling its immediate tools. Anything that would take longer than a few seconds is handed off to a background bot, a text model from OpenAI, Anthropic, Google or xAI, equipped with a shell, a real browser, the user's files, Agent Skills and MCP servers, so the voice call can keep talking instead of going silent.
A job handed to one bot can bring in others, hold a step until a needed result is in, check what comes back, and ask the user only when a decision is genuinely theirs; every handoff is saved so a user can open the thread to see who did what. Results such as pages, charts, videos, slides, documents and scripts are saved as files, and scheduled routines can start jobs on a timer.
A bot can operate its own browser or the user's already-signed-in Chrome, stopping short of pressing a pay button so the user completes purchases themselves. Users can also message the assistant through Telegram, Discord or Slack instead of talking. It requires Node.js 22.18 or newer and one OpenAI API key, and is written in TypeScript.