Azure AI logo Microsoft Foundry

Magentic-UI

Unlocking the full potential of AI requires the development of effective mechanisms for human-AI collaboration. By reducing cognitive load while ensuring users remain in control, AI can significantly enhance human capabilities and streamline complex workflows. Magentic-UI was designed with this goal in mind, serving as a research platform aimed at advancing research on human-in-the-loop experiences.

Open Magnetic-UI Repo Released for research purposes.

Key features of the Magentic-UI research include:

  1. Co-planning. Magentic-UI allows users to directly modify its plan through a plan editor or by providing textual feedback before Magentic-UI executes any actions.
  2. Co-tasking. Users can pause the system and give feedback in natural language or demonstrate it by directly taking control of the browser.
  3. Action guards. Magentic-UI seeks user approval before executing potentially irreversible actions, and the user can specify how often Magentic-UI needs approvals. Furthermore, Magentic-UI is sandboxed for the safe operation of tools such as browsers and code executors.
  4. Task learning. Magentic-UI can learn and save plans from previous interactions to improve task completion for future tasks.
  5. Sentinel Steps. Magentic-UI can now handle long-running monitoring tasks (e.g. tracking prices or repo updates) and repeatable workflows (e.g. updating a document on a weekly basis).