Use the Jupyter AI Agent (Tutorial)—DRAFT
Last updated 2026-07-27At a Glance
Jupyter AI Chat lets you ask SDK‑aware questions in your notebook and get grounded, runnable answers without leaving the IDE. You can also use the agent to generate, clean up, and validate notebooks based on internal best practices and example workflows.
When to Use This Feature
Use Jupyter AI chat when you want AI help that’s aware of your SDKs, notebooks, and internal docs, instead of generic code suggestions. Here are some examples:
You need to understand or apply HISE SDKs without digging through the documentation.
You want the agent to turn a natural‑language description of an analysis into structured notebook cells, comments, and code.
You’re cleaning up legacy notebooks and want help making them compliant, reproducible, and easier for others to use.
You have questions like “How do I upload files?” or “How do I save a visualization?” and want to get answers directly in the notebook.
Your team wants to create custom personas that capture shared workflows so everyone can ask the same agent for consistent guidance.
NOTE
Lorem ipsum
Instructions
Select and prompt the agent
Log in to HISE, go to Research > IDEs, and opent an IDE.
Find the agent selector in the chat panel, and choose the agent that best matches your task: for example, ask B cell for general IDE help or Claude Code for coding help
In the chat input, describe what you’re trying to do in natural language, such as “Upload results to HISE from this notebook” or “Explain how to use the uploadFiles function.”
A. Optionally paste relevant cells or snippets so the agent can ground its answer on the exact code you’re using.
B. To generate or transform notebooks with the agent, use prompts like “Given this notebook, write a commented version to /home/workspace/mycommentednotebook.ipynb” or “Create a new notebook that loads data, runs SCVI, and saves results back to the platform.”
C. Let the agent produce notebook cells, comments, and metadata using internal examples and templates from the/examplesfolder, Drive, or your own files.
D. Open the generated notebook, run it end‑to‑end, and make any edits needed for your specific dataset or workflow.