Developers introduced the Panel workspace for AI agents

Developers have released Panel — a research environment in which an AI agent interacts with the user in real-time. The tool allows the agent to independently create custom control panels, widgets, and applications for data visualization directly in the workspace.

The system is focused on working with files, PDF documents, and notebooks. The agent manages the interface structure, placing the necessary tools in the dock panel as tasks are performed. This allows the user not to switch between tabs, as the agent dynamically adapts the environment to the current research context.

The project is published in open access on GitHub under a license that allows local deployment. The tool is designed to automate routine operations when analyzing large data sets, where constant visualization of intermediate results is required. Users can integrate their own models to manage the behavior of the agent within the environment.

The development is aimed at changing the interaction with AI agents, turning them from chatbots into active participants in the workflow. Unlike standard interfaces, where the agent is limited to a text field, Panel gives the model the ability to independently form a visual layer for displaying answers and managing the file system.