Notebook functionality within Orchestrate UI

Inspired by the release of Open Notebook, similar functionality within the Orchestrate environment would be powerful, especially if it can connect to WatsonX.ai-hosted models, and/or use the Orchestrate AI Gateway.

The idea would be to create a private research and personal-productivity mechanism akin to NotebookLM but with the privacy, agentic features and controls that come with Orchestrate. Notebooks could be saved and retained for subsequent editing, sharing (within the tenancy), export (content-only linked or exported as PDF etc.). The API could also enable Notebook content to be extracted and posted to a client EDRMS as a finished document.

The idea could be extended with agentic connections to databases / unstructured data sources within the client's network - enabling papers, presentations, documentation and even Jupyter-style code, SQL or even analytics embeddings. An MCP client - the Orchestrate toolkit mechanism - could enable requests to be made within a Notebook against external MCP-enabled knowledge and agent sources.

An instance might include an admin-maintained whitelist of available services (including deep research agents crafted within Langflow / Orchestrate) and knowledge, as well as a choice of models through the AI gateway. Consideration would need to be given to access controls / security, as well as logging and asset (i.e. notebook / content) retention and visibility.