Can this do preliminary "Deep Research"?

[TomLucidor]

I am trying to find alternatives to EmergentMind.com, TheMoonlight.io, and AIModels.FYI, when it comes to paper-hunting (AI Scientist uses OpenAlex) + related document graphing + "Explain like I am 14" summarization + glossary/keyword capture + knowledge gaps or open problems. Which parts can be replaced with Open Notebook?

[lfnovo]

Moving this from Q&A to Ideas and keeping it open as a product conversation about academic deep-research workflows.

Open Notebook can currently ingest papers, search and ask questions over supplied sources, and generate grounded summaries or notes. It does not yet provide the complete workflow described here: native paper discovery through services such as OpenAlex, related-document graphing, autonomous research, or systematic identification of knowledge gaps.

Related design work exists in #319 for a built-in research agent and #973 for a web-research agent. This Discussion is the right place to contribute academic use cases, expected outputs, trust requirements, and boundaries between source discovery, analysis, and autonomous research.

[TomLucidor]

Feels like this can be a general hub somewhere between Row-Bot + TLDW + Emulo + similar apps, and deeper academic research workflows + comprehensive caching mechanisms.
Even within document graph, we need core topic vs subtopics vs adjacent/bridge topics, as well as keywords/concepts as isolated parts. Some kind of graph structure should be well-articulated such that lateral thinking is promoted, and use as little tokens as possible.

[lfnovo]

Status refresh: what this thread asked for is the deep research mode in #1315: iterative search, opening sources, expanding, comparing — sharing the same evidence contract as the simple mode, so citations are verifiable in both. Questions 4 and 5 there (what to show while the agent works; what to call the modes) are the ones your experience answers.

[TomLucidor]

Include considerations on conflicts and knowledge gaps and doubts/distrust as well.