How to Setup Claude Skills in Gemini + Google Notebooks!
The video demonstrates how to use Google's Notebook LM (powered by Gemini) to generate Claude skill files (skill.md) without hallucinations, by grounding outputs in user-supplied sources. The presenter walks through a five-step process: creating a notebook, adding focused sources, generating the skill file, importing it into Claude, and testing it. The method is positioned as a way to build a scalable 'AI agency' of stacked skills.
Summary
The video, presented by a digital avatar representing Julian Goldie of Goldie Agency, introduces a workflow for creating Claude AI skills using Google's Notebook LM tool powered by Gemini. The core concept revolves around 'Claude skills,' which are structured instruction files (skill.md) stored in folders that tell Claude exactly how to perform a specific task — eliminating guessing, hallucinations, and inconsistent outputs.
The presenter identifies the core problem: building Claude skill files from scratch is time-consuming and error-prone. The proposed solution is to use Notebook LM as a research and generation engine. Because Notebook LM answers questions based only on user-supplied sources (PDFs, websites, YouTube videos, documents), the outputs are grounded in real information rather than fabricated content.
The five-step workflow is outlined as follows. Step one involves opening Notebook LM at notebooklm.google.com, signing in with a Google account, and creating a new notebook. Step two requires adding high-quality, focused sources relevant to the skill being built — the presenter uses writing high-converting landing pages as the example, suggesting sources like copywriting guides, expert YouTube videos, case studies, and Anthropic's official Claude skills documentation. Step three is the research and generation phase, where the user prompts Notebook LM to synthesize the sources into a structured skill.md file using a specific prompt template provided in the video. Step four involves moving the generated file into Claude Code's skills folder under a descriptively named subfolder. Step five is testing the skill by asking Claude to perform the target task and iterating based on results.
The presenter also shares several 'pro moves': organizing sources by topic, using one notebook per skill, keeping each skill focused on a single job, using Claude itself to refine the skill after Notebook LM generates it (described as a 'double loop' where Gemini researches and Claude refines), and version-controlling skill files like software products.
A practical example is given for building a welcome email skill for an AI community, demonstrating how the same process applies across use cases like cold email writing, SEO content, customer support, sales scripts, and product descriptions. The presenter frames stacking multiple skills as building a 'mini AI team' capable of running significant business operations without writing any code.
The video closes with safety advice — always human-reviewing skill files before deploying them — and a call to action for the presenter's paid community (AI Profit Boardroom) and free community (AI Success Lab).
Key Insights
- The presenter argues that Notebook LM eliminates hallucinations in Claude skill files because it answers questions based exclusively on user-supplied sources rather than its general training data, making the generated skill.md files grounded in real, verifiable information.
- The presenter describes a 'double loop' process where Gemini (via Notebook LM) handles the research and initial skill generation, and then Claude itself is used to further refine, tighten, and add examples to the skill file — combining the strengths of both AI systems.
- The presenter advocates for keeping each Claude skill narrowly scoped to a single job, warning that attempting to build a 'mega skill' that handles 10 tasks at once confuses Claude and degrades output quality.
- The presenter provides a specific reusable prompt template for generating skill files in Notebook LM, instructing it to create a structured skill.md that includes defined sections (headline, subheadline, value proposition, social proof, CTA, FAQ) so Claude can follow it consistently every time.
- The presenter frames stacking 10 or more specialized skills as equivalent to running a 'mini AI team' capable of handling email writing, landing pages, social posts, customer replies, and SEO content — all without writing a single line of code.
Topics
Transcript
[0:00] How to set up Claude skills in Gemini plus Google notebooks? Sign up. Today I'm going to show you something crazy. You can now turn Google's notebook LM into a Claude skills factory. This is huge. It's new. And it's stupid easy. In the next few minutes, you'll learn how to build custom AI skills in minutes, not hours. And the best part? Cuts hallucinations like a hot knife through butter. Let's go. So let's break this down. What are Claude skills? Why does this matter? And how do you actually set this up? Stick with me. Cuz by the end of this video, you'll have a step-by-step system to build your own [0:30] Claude skills using Gemini and…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Julian Goldie SEO
NEW Nvidia Autonomous AI is WILD!! 🤯
Nvidia announced Nemo Clo, a new autonomous AI agent system that operates independently without continuous prompting. Powered by Nemotron 3 Ultra (a 550 billion parameter model), the system is five times faster and cheaper than previous versions, with OpenShell providing secure sandboxed execution.
Laguna XS 2.1: New FREE + Opensource Local AI!
Julian reviews Laguna XS 2.1, a new free open-source local AI coding model from Poolside that performs comparably to Qwen 3.6 and outperforms Claude Haiku on benchmarks. He demonstrates its practical capabilities by building landing pages and functional apps, highlighting its speed, offline functionality, and multiple deployment options through local setup, Claude Code, or OpenRouter's free API.
How to Run Hermes FREE Forever!
The video demonstrates how to run the Hermes AI agent for free using Gemma 4, a local open-source model from Google, with significant speed improvements through MLX optimization. The setup works on Apple Silicon Macs or via free APIs on Open Router, enabling autonomous agents to work offline and privately without subscription costs.
This NEW Chinese AI is INSANE! (FREE + Open Source!)
Long Cap 2.0 is a new open-source Chinese AI model from a food delivery app company that offers 1 million tokens of free context memory, beats GPT-4.5 on SWE bench pro benchmarks, and uses efficient parameter activation to reduce computational overhead while maintaining high performance.
Claude Code is now FREE: Here’s how…
Google's new Gemma 4 model running on Ollama is 90% faster on Apple Silicon, enabling free Claude Code usage locally without token costs. The setup requires three simple steps: downloading Ollama, Gemma 4, and installing into Claude Code, with alternatives available via OpenRouter API for non-Mac users.