PyCharm Integrates Live Jupyter Kernel for AI Agents

JetBrains has introduced a groundbreaking feature in PyCharm: AI agents can now operate directly within a live Jupyter kernel. This innovation addresses common frustrations such as corrupted .ipynb files, lost model states, and inefficient token usage during long jobs. By allowing AI agents to work within the kernel, state persistence across cells is maintained, and idle checking is minimized, reducing costs by up to 12% in tested scenarios. The new Jupyter skill leverages PyCharm’s notebook intelligence, enabling agents to execute Python code seamlessly, wait for cell completion, and read incremental output, all while keeping prompts cache warm. This functionality is available to users with a JetBrains AI subscription, promising enhanced efficiency for machine learning workflows.