Show HN: Panel – A research workspace where the agent can build its own panes (github.com)

A new research workspace called "Panel" has been announced, designed to enhance collaboration between AI agents and researchers by integrating chat, file management, PDFs, and Jupyter notebooks into a single interface. In its early build, users can create custom viewers and applications tailored to their workflow, allowing AI agents to read, write, and interact with files while engaging in conversations. Key features include support for long-running commands, customizable panes, and a structured workspace that maintains its layout across sessions, which is critical for researchers who often juggle multiple file types. The significance of Panel lies in its ability to streamline research processes by enabling AI to assist in literature reviews and provide real-time performance updates on tasks. This workspace utilizes a robust set of modules—akin to AI skills—but with enhanced input/output definitions and the capability to handle complex inter-module workflows, promoting clearer communication and validation during module execution. With a data abstraction layer bridging in-memory and filesystem objects, developers can create versatile applications that leverage both local files and in-memory data, making Panel a powerful tool for the AI/ML community aiming to facilitate research efficiency and transparency.