Claude Code + NotebookLM + Obsidian: The Research Stack Nobody's Using

Artem Zhutov ยท Intermediate ยท๐Ÿ“„ Research Papers Explained ยท1mo ago
๐Ÿ“š Obsidian x Claude Code Lab: https://lab.artemzhutov.com โ†’ Build your own AI memory system, vault architecture, and agent skills - live, with me ๐Ÿ’ฌ Discord: https://discord.gg/g5Z4Wk2fDk Claude Code can't read 300 files at once. Context window isn't big enough. So I put it inside NotebookLM - it reads everything, cites every source, and Claude Code controls it from the terminal. I show the full workflow: YouTube search to sources, cited answers with passage-level deep links back to Obsidian, audio overviews you can listen to on your phone, and the full loop where notes go in and a knowledgeโ€ฆ
Watch on YouTube โ†— (saves to browser)

Chapters (24)

300 Files Don't Fit in Your Context Window
1:36 What Is NotebookLM
2:30 Claude Code + notebooklm-py CLI
3:07 How It All Connects (Architecture Diagram)
3:58 Live Demo: Researching a Topic from Terminal
5:00 YouTube Search (20 Videos Found)
5:57 Creating a Notebook + Adding Sources
6:32 Sources Loading in Real Time
7:14 Dashboard with 20 Sources in Obsidian
7:56 Importing Sources as Vault Files
8:55 First Q&A: Gaps and Opportunities Across 20 Videos
9:16 Verifying Citations Against Source Videos
11:17 Citation Accuracy Audit (60% Strong Match)
11:45 Graph View: Connections Between Sources
12:04 Building an Obsidian Base for NotebookLM Sources
13:06 Topics Extracted as Wikilinks
14:12 Obsidian x Claude Code Lab
15:31 Obsidian CLI Workshop
16:34 Adding More Sources (Articles, PDFs, Vault Notes)
18:23 NotebookLM Supports Up to 300 Sources
18:50 Custom Persona per Notebook
19:54 Q&A with Citation Resolution in Obsidian
21:03 Passage-Level Deep Links (Click to Exact Chunk)
21:54 Before vs After: 20 Tab
Recovering the Humankind's Past and Saving the Universal Heritage
Next Up
Recovering the Humankind's Past and Saving the Universal Heritage
Coursera