How to Use Google Gemini Gems with Notebook LM for Content

01What Google Gemini Gems Are and How They Work

Google Gemini Gems are custom versions of Gemini that remember your rules forever. Think of a gem as a smart helper that knows exactly who to be, what to do, and how to format every answer. You set the rules once, and the gem follows them for every single request.

Julian Goldie explains that gems fix a big problem with regular AI chats. In a normal Gemini chat, your rules start to fade after a few messages. The AI drifts away from your tone and format.

But a gem locks in your rules for life. You never have to repeat yourself.

To find gems, open Gemini and click Gems on the left side. Then click My Gems and hit New Gem. You will see three boxes: Name, Description, and Instructions.

The name is just for you. The description helps you remember what each gem does. The instructions box is where the magic happens.

02The PACT System for Writing Gem Instructions

Julian Goldie uses a simple system called PACT to write gem instructions. PACT stands for Persona, Assignment, Context, and Template. This system makes sure your gem always writes the way you want.

Here is how PACT works with a real example. Say you run a coaching group called the AI Profit Boardroom and you want welcome emails.

• Persona: Write who the AI should be. Example: "You are a friendly coach who helps business owners use AI to save time."
• Assignment: Tell the AI exactly what to do.
• Context: Give extra rules. Example: "Always end with one clear next step. Match a tone that is warm but not too casual."
• Template: Show the format you want.

Once you save these instructions, you can ask the gem to write 50 emails and every single one will follow those same rules. Short. Clear.

With a next step. No drift. No repetition.

03How to Set Up Gemini Gems and Notebook LM Together

1. Open Notebook LM and create a new notebook. Give it a name like "Business Knowledge Base."
2. Upload your source files into the notebook. These can be reports, blog posts, case studies, policies, or brand voice guides.
3. Switch the prompt box in Notebook LM to deep research. Type a question like "Find more sources on how small business owners use AI automation to grow in 2026."
4. Let Notebook LM dig through sources. When done, click import to add the new sources to your notebook.
5. Go to the Studio panel on the right. Click Reports and pick Create My Own. Notebook LM gives you a clean research brief with every claim linked to a real source.
6. Export the research brief to Google Docs. It saves to your Drive automatically.
7. Open Gemini and create a new gem using the PACT system. Name it something like "Content Writer" and write your full voice, tone, and format rules in the instructions.
8. Do not upload any specific research to the gem. The gem only holds your style so you can use it for any topic.
9. Open Google Docs. Click the Gemini icon in the top right. Turn on Drive search so the gem can read your files. Select your gem and type a prompt that tells the gem to read the research brief in your Drive and write the content.

04Three Systems You Can Build with This Setup

Julian Goldie shows three practical systems you can build this weekend. Each one takes about 20 minutes to set up and saves hours of work.

System 1: Content Writer. Before this system, Julian spent an hour reading articles, pulling stats, and writing from scratch. Now he opens Notebook LM, uploads a few sources like a report on AI tools, and runs deep research.

He exports a research brief to Google Docs. Then he uses his Content Writer gem to pull the brief from Drive and write a full blog post in his voice. The whole thing takes 15 minutes.

System 2: Proposal Writer. For service businesses, every new lead needs a proposal. Julian uploads case studies and service descriptions to Notebook LM.

He generates a master reference doc. Then his Proposal Writer gem reads that doc and drafts a full proposal with project scope, timeline, and matching case studies. It pulls real projects, not made-up ones.

System 3: Support Helper. Customer support questions repeat every day. Julian uploads his return policy, shipping policy, subscription terms, and brand voice guide to Notebook LM.

He generates a briefing doc and a response guide. Then his Support Helper gem reads those docs and drafts two replies for any customer question: a full email and a short chat reply. Same rules every time.

05Using the Content Writer System Step by Step

Julian Goldie walks through the content writer system in detail. This is the first system he built and the one he uses most often.

• Open Notebook LM and create a new notebook. Upload sources like a report on AI tools for small business or blog posts about AI automation trends.
• Switch to deep research mode. Type "Find more sources on how small business owners use AI automation to grow in 2026." Notebook LM finds and imports relevant sources.
• Go to the Studio panel. Click Reports and pick Create My Own. Notebook LM builds a research brief where every claim links back to a real source. You can click any claim to verify it.
• Export the research brief to Google Docs. It saves to your Google Drive.
• Open Google Docs. Click the Gemini icon in the top right. The sidebar opens. Click the plus icon and turn on Drive search so the gem can read your files.
• Pick your Content Writer gem. Type a prompt like "There is a research brief in my Drive called AI Automation Trends 2026. Read it and write a 1200 word blog post for the AI Profit Boardroom. Show how AI automation helps small business owners save time and get more leads. Follow my voice and structure."
• The gem pulls the report from Drive, reads the real research, and writes a full blog post in your exact voice. The whole process takes about 15 minutes.

06Results: Speed and Accuracy You Can Trust

Before this setup, Julian Goldie spent an hour writing a single piece of content. He had to read articles, pull stats, take notes, open a blank doc, and start from zero. The research never connected to the writing.

He had to read his own notes again and try to remember which stat came from where.

Now the same task takes 15 minutes. The gem writes in his voice, and Notebook LM makes sure every fact is real. Every claim in the research brief links back to a real source.

Julian can check anything he wants with one click.

Julian also shows that without Notebook LM in the mix, the gem would guess at details and make up case studies. With the pre-organized reference, the gem pulls real projects and real facts. The same system works for proposals and customer support.

A proposal that used to take an hour now takes 15 minutes. Customer support replies stay consistent across the whole team because the gem knows every policy.

07Tips, Limits, and Mistakes to Avoid

Gems have one weak spot. They follow your rules well, but when you ask them to use real facts from your files, they sometimes guess. They paraphrase loosely or fill in the gaps with stuff that sounds right but is not real. That is why you must pair gems with Notebook LM.

Notebook LM is built to stick to your sources and never makes things up.

Do not upload specific research to the gem. The gem should only hold your style, voice, and format rules. Keep your research in Notebook LM and your Drive.

This way you can reuse the gem for any topic without changing anything.

Always turn on Drive search when using the gem in Google Docs. Without it, the gem cannot read your files and will fall back to guessing. Also, make sure your reference docs are organized and saved in Drive before you start.

If the gem cannot find the file, the whole system breaks.

New team members can use the same gem to write in your voice without training. But they still need access to the Drive files and the right gem. Set that up before you hand it over.

08Who This Setup Is For and Why It Matters

This free setup is for anyone who creates content, writes proposals, or handles customer support. If you run a service business, a coaching group, or any company that sends emails, proposals, or support replies, this will save you hours every week.

Julian Goldie explains the bigger win: you stop wasting time on stuff a machine can do. You spend that time on things that actually grow your business. One gem becomes a full company support system.

New team members can be ready in a day instead of weeks because the gem already knows every policy and every rule.

Pick one of the three systems—content writer, proposal writer, or support helper. Build it this weekend. It takes about 20 minutes.

Once you have one working, you will see how easy the others are to set up. The whole point is to let the AI handle the repetitive work so you can focus on growth.

09What to Do Next with Your New Systems

Julian Goldie recommends picking one system and building it this weekend. The content writer system is the easiest to start with because it only needs a few source files and one gem.

After you build the first system, share the gem with your team. Anyone on your team can use the same gem to write proposals, support replies, or blog posts using the same rules and the same facts. You do not have to be involved.

If you create a lot of content but nobody finds it online, that is a separate problem. Julian mentions that his team can help with SEO to get your content ranking on Google. But the core of this episode is the gem and Notebook LM system.

Build that first. Once your content is fast and accurate, then worry about traffic.