Currently, GoodNotes AI can only be used for basic document search (an advanced version of search) and asking questions within a document. Even when hints or correct answers exist in the document, it often responds with messages like, “The requested content cannot be found in the document. I will answer using general knowledge.”

In addition, I would like the AI to provide answers that directly address my questions, but instead it often gives summarized or reorganized responses, making it difficult to find the exact answer I am looking for.
For example, when I ask, “Is XX correct?”, the response usually explains “XX is …” by summarizing related content about XX. In these cases, it is hard to clearly determine whether XX is correct or incorrect.

The GoodNotes AI performance I envision includes the following:

In-document AI chat (answers strictly based on the document) / Out-of-document AI chat (answers using general or prior knowledge)

Reasoning ability (the ability to reason and make judgments based on the given document)

File upload support, including images, PDFs, PPTs, Word files, and native GoodNotes files

Web search functionality:

In-document AI chat + web search (if the requested information is not found in the document, use additional web search)

Out-of-document AI chat + web search (use web search to supplement answers)

Deep research capabilities

The following features would be great to have after performance improvements:

Image generation

Video generation

Features similar to those in Google LM (such as AI audio overviews, PPT generation, etc.)