Goal:
Introduce the "Notebooks" feature in Developer Lightspeed, providing users with isolated workspaces to upload project-specific documents and interact with an AI that generates highly accurate, citation-backed answers based solely on the content of these documents. This feature aims to enhance context-specific troubleshooting, onboarding, and project reviews through RAG-based (Retrieval Augmented Generation) interactions.

Scope:
Ensure the seamless integration of the Notebooks feature, including document upload, context isolation, and RAG chat. This includes UI enhancements, backend integration with vector databases, and enforcing proper data privacy controls. The feature will also support RBAC for controlled access and will allow admins to enable/disable the feature through configuration.

Acceptance Criteria:

• Users can easily create, manage, and delete notebooks from a dedicated UI section.

• Supported file formats for upload include PDF, Markdown, Office Docs (Word, Excel, PowerPoint), with size and type restrictions enforced.

• The RAG chat model uses only the uploaded documents for answering queries, with clear "Sources" chips indicating which files were referenced in responses.

• The UI is updated to reflect the new notebook functionality, including a new tab for notebooks and refined chat and history panels.

• Admins can enable the Notebooks feature via configuration flags and manage permissions through RBAC policies.

• File upload is constrained to ensure system stability (e.g., max file size and number of tokens), with error messages for unsupported formats.

Out of Scope:

• Generative content features like study guides, podcasts, or content generation beyond Q&A.

• Web scraping or URL ingestion—only file uploads are supported.

• File types like scanned PDFs, images, audio, and video are not supported.

• Sharing of notebooks across users or teams is not part of the initial release.

Done Checklist:

• All E2E tests pass in CI, ensuring notebook creation, file uploads, and RAG-based chat functionality work as expected.

• UI elements (Notebook creation, file upload, chat interactions) are fully implemented and aligned with Figma designs.

• The system correctly isolates data between notebooks, ensuring users can only access their own documents.

• Admin configuration works, enabling/disabling the feature and managing RBAC permissions.

• The RAG chat returns responses based only on the documents uploaded within the notebook, with proper source citations.

• File upload functionality enforces size and format constraints, with clear error handling in place for unsupported files.