AI Writing, Technical Writing
Using Google NotebookLM to Write FAQs – A Klariti Primer
Jul
I’ve tried a lot of tools to gather information, from Notion to Obsidian, but I’ve found that Google’s NotebookLM helps me gather material in such a way that I can write more useful FAQs. It’s pretty intuitive to use.
I’d like to walk you through how I use it for technical documentation. Hopefully, this will be of use to you when updating your document set. It’s also a nice way to get into using AI tools in a practical way.
So, what makes it so different?
NotebookLM has a fascinating feature that changes the game for creating accurate content. It’s built on a principle called Retrieval-Augmented Generation, or RAG.
That sounds technical, I know. But the idea is simple. Think of it as giving an AI an open-book exam.
A general-purpose chatbot like Gemini or ChatGPT pulls answers from its vast, public training data—essentially its “memory” of the internet. It’s guessing. But with NotebookLM, you first give it your own specific documents—your “textbooks.” The AI is then forced to find answers only within the material you provided. It can’t make things up. This single shift makes it a powerful ally for factual work.
Let’s be honest. I personally believe that AI will never replace the role of a subject matter expert or someone who has lots of niche domain knowledge. It lacks the human empathy required to read between the lines and pick up on the cues from customers.
However, you can use it as a deluxe co-writer. It’s the tool that handles the heavy lifting, so you can focus on nuance, planning, and strategy.
FAQ Guide Templates – Instant Download
Here’s my workflow.
1. Add Your Sources
Everything starts with the quality of your source material. NotebookLM is only as smart as the documents you give it. Before I do anything else, I create a new “notebook” and feed it the raw intelligence.
What do I add?
- A dump of the 50 most recent, relevant support tickets.
- Transcripts from customer interviews.
- The existing product user manual.
- Technical specification documents from the engineering team.
- Marketing brochures and website copy.
You can upload PDFs, text files, or just copy and paste from Google Docs. The goal is to give the AI a 360-degree view of your product and your customers’ current reality.
2. Generate Foundational FAQs
Once my sources are loaded, I start with a simple prompt. I don’t ask it to write the final FAQ yet. I ask it to think like a customer.
My prompt:
“Acting as a brand new user who has just purchased our product, review all the provided sources and generate a list of the 20 most likely questions you would have.”
In seconds, NotebookLM will produce a list. Some questions will be obvious, but I guarantee a few will surprise you. This gives me an instant, data-driven outline of what the FAQ page needs to cover.
3. Identify Gaps in Your Documentation
This is where I find it really helpfu. Your FAQs should answer user questions, but they can also reveal holes in your primary documentation.
My prompt:
“Based on all the provided sources, what important information appears to be missing? What questions might a user have that cannot be answered by the current documents?”
NotebookLM will compare the user questions from the support tickets against the answers available in your user manual. It might point out something like, “The manual explains how to export data, but multiple support tickets ask why the exported CSV file is formatted a certain way. This ‘why’ is not explained.”
You’ve just found a critical documentation gap to fix. It”s super hard to do this if you have 100s of FAQs you need to keep on top of.
4. Research New FAQs
When NotebookLM generates a question, it also provides citations, showing you exactly which of your source documents it used. This is incredible for research.
If it suggests the question, “How do I integrate with Salesforce?”, it will show [1, 4, 5] next to it. I can click those numbers and instantly see the snippets from the support ticket, the marketing page, and the tech spec that mentioned Salesforce. This saves me hours of searching for context.
FAQ Guide Templates – Instant Download
5. Write the Answers
Now I start drafting. I take a question generated in step two, and I use the research from step four to build the answer. This is a collaborative dance.
My prompt:
“Using only the provided sources, write a clear, step-by-step answer to the question: ‘How do I reset my password?'”
It produces a clean, factual first draft. My job is to take that draft and elevate it. I rewrite it for tone, add helpful context, and make sure it sounds human.
6. Edit with Precision
The AI’s first draft is the starting block, not the finish line.
Editing is a human-centered task. I treat it like a Formula 1 team doing a systems check before a race—every detail must be perfect.
I review the AI-generated answer side-by-side with the source documents it cited. Is it 100% accurate? Did it misinterpret a technical term? Did it capture the spirit of the solution, or just the literal steps? This is where my expertise adds the real value.
7. Update with Kaizen
Your FAQ page is a living document. It needs constant care. I’m a fan of the Japanese concept of Kaizen, or continuous improvement.
Each month, I add a new batch of support tickets to my notebook. Then I ask it a simple question:
“What new questions have appeared in the latest sources that are not in our original list?”
This process keeps the FAQ page fresh and prevents it from becoming a digital relic.
8. Create On-Demand Audio Overviews
Some people find this gimmicky but I use it a lot. NotebookLM can generate an audio summary of your notes or a selection of text. I use this to create quick, custom podcasts for myself and our team.
After I’ve finalized a new section of five FAQs, I select them and prompt: “Create a 3-minute audio overview of these new Q&As, explaining the problem and the solution for each.”
I then share this MP3 file in Slack or Teams. Now, our support agents can get up to speed on the latest documentation updates while making a coffee. It’s faster and more engaging than asking them to read another document.
This entire process, from research to writing to the final podcast, is about using AI to augment your own expertise. It’s an endurance event, and NotebookLM is the tool that helps you keep a steady pace, conserving your expert energy for the parts of the race where it matters most.
So, I hope you found that useful. Make sure to connect on the socials and get the newsletter for special offers and updates.
Get Templates – Instant Download:
Connect with us
