Sponsored by
The World's Biggest Dev Event Hits Silicon Valley
WeAreDevelopers World Congress comes to San José, CA — September 23–25, 2026, in the heart of Silicon Valley. 10,000+ developers, 500+ speakers, 20+ stages, and the full software development lifecycle in one place.
Kelsey Hightower. Thomas Dohmke (fmr. CEO, GitHub). Christine Yen (CEO, Honeycomb). Mathias Biilmann (CEO, Netlify). Olivier Pomel (CEO, Datadog). The people actually building the tools you use every day — all on one stage.
Three days of AI, agents, cloud, security, and architecture, plus workshops, live coding, and the official Congress party. Prices rise as the event gets closer — lock in today's rate.
Use code GITPUSH26 for 10% off.
Welcome to Grind Engineer, your guide to becoming a better software engineer! No fluff. Pure engineering insights.
Most engineers read 50+ articles a month and remember almost nothing. The problem isn't effort. It's that reading without a system is just entertainment with extra steps.
Three tools, wired together the right way, change everything: Google's NotebookLM for source grounded research, Claude Code for orchestration and processing, and Obsidian for permanent storage. Each one is good alone. Together they create a learning loop that compounds over time.
Here's what the full stack looks like:
What Each Tool Actually Does
Before we wire them together, let's get clear on what each piece brings to the table.
Tool | What It Does | Why It Matters |
|---|---|---|
NotebookLM | RAG powered Q&A grounded in YOUR sources, with citations | No hallucinations. Every answer points back to your documents |
Claude Code | AI coding agent that runs skills, calls MCP tools, manages files | The glue layer. It reads, writes, and orchestrates everything |
Obsidian | Local first markdown knowledge base with backlinks and graph view | Your files. Your disk. No vendor lock in. Search across everything |
NotebookLM deserves a closer look because most people underestimate it. It's not just another chatbot wrapper. You can load up to 50 sources per notebook: PDFs, YouTube videos, EPUBs, slide decks, web URLs. Then every response it gives you is cited back to those specific sources.
The killer feature? Audio Overviews. NotebookLM generates podcast style conversations about your sources. Two AI hosts discuss your material, and with Interactive Mode, you can literally join the conversation. Pause the hosts. Ask a question. They answer using your sources. It's wild.
The Workflow (Step by Step)
Here's how the three tools connect in practice:
Step 1: Feed sources into NotebookLM
Drop in the papers, docs, videos, and articles you want to learn from. Create a notebook per topic. "Distributed Systems" gets its own notebook. "System Design Patterns" gets another.
Step 2: Research and extract with NotebookLM
Ask questions. Generate summaries. Create flashcards. Use the Audio Overview to hear concepts explained in a conversational format. NotebookLM grounds every answer in your sources, so you know exactly where each fact came from.
Step 3: Claude Code processes and structures the output
This is where it gets interesting. Claude Code takes your raw research and transforms it into structured markdown. It can:
Convert messy notes into clean Obsidian compatible files
Add YAML frontmatter (tags, date, source links)
Create backlinks to related concepts automatically
Generate summary tables and comparison matrices
Step 4: Output lands in your Obsidian vault
Every processed note drops into your vault as a .md file. Tags connect it to related topics. Backlinks let you navigate between concepts. The graph view shows you how everything relates.
💡 Key Insight: The real power isn't in any single tool. It's that every research session makes your vault smarter, and Claude Code can read past notes to build on what you already know.
Why This Compounds
Here's the part that makes this stack different from just "using AI to take notes."
Say you spend Tuesday researching distributed consensus. Your vault now has structured notes on Paxos, Raft, and leader election. Three weeks later, you're studying database internals. When Claude Code processes your new research, it can reference your existing notes on consensus protocols.
Your vault becomes a personal knowledge graph that grows with every session.
Session | What Happens | Vault State |
|---|---|---|
Week 1 | Research CAP theorem | 5 connected notes |
Week 3 | Study database replication | 12 notes, backlinks to CAP theorem notes |
Week 6 | Prep for system design interview | 25+ notes, Claude Code synthesizes across all of them |
The compounding effect is real. Engineers who've adopted this workflow report spending 60% less time re researching topics they've already covered.
What Makes This Better Than [Other Tool]
You might be thinking: "I already use Notion" or "ChatGPT does this fine."
Here's the difference:
Approach | Grounded Citations | Local Files | Compounding Knowledge | Orchestration |
|---|---|---|---|---|
ChatGPT alone | ❌ | ❌ | ❌ | ❌ |
Notion AI | ❌ | ❌ | Partial | ❌ |
NotebookLM alone | ✅ | ❌ | ❌ | ❌ |
This stack | ✅ | ✅ | ✅ | ✅ |
ChatGPT hallucinates. Notion locks your data in their cloud. NotebookLM alone doesn't connect sessions together. This stack solves all three problems.
A Real Example
Let's say you want to deeply understand how Netflix handles failover.
NotebookLM: Load 5 Netflix engineering blog posts, 2 conference talks (YouTube URLs), and the relevant chapters from "Designing Data Intensive Applications"
Ask NotebookLM: "How does Netflix handle regional failover? Cite specific sources." You get a grounded answer with page numbers and timestamps
Claude Code: Takes that output, creates
netflix-failover.mdwith proper frontmatter, backlinks to your existing notes on[[load-balancing]]and[[circuit-breakers]], and a summary tableObsidian: The note appears in your vault. Graph view now shows Netflix failover connected to your broader distributed systems knowledge
Total time: 20 minutes. Quality of retention: dramatically higher than just reading the blog posts.
Try This Today
Create your first NotebookLM notebook. Pick a topic you're actively studying. Add 5 to 10 sources (blog posts, papers, YouTube talks). Ask it 3 questions and notice how every answer cites your sources directly.
Set up an Obsidian vault for engineering knowledge. Create folders for your core areas: system design, language internals, infrastructure. Start with one processed note using the YAML frontmatter pattern:
tags,date,sources,related.Connect them with Claude Code. Ask Claude Code to take your NotebookLM research output and structure it as an Obsidian compatible markdown file with backlinks and tags. Do this once and you'll immediately see the value of the full pipeline.
Sources
See you in the next one!
Scortier, Signing Off!