top of page

The making of... an AI supported blog post: using notebookLM in practice in a rather unsophisticated way

My blog unexpectedly has found more than 3 readers.

And only Africa seems to be the "white continent".

So, I should explain how I use AI responsibly, as a nearly 70 year old. Here, I will explain my use of Google´s free NotebookLM.

I assume that process is just basic and primitive stuff for many, way too under-automated.

But for some reader may be interesting, if you like research but haven´t used NotebookLM or a similar tool (there are quite a few - but they cost money).

About NotebookLM

Hey I am not going to re-invent the wheel by explaining notebookLM...I was made aware through this guy. He fascinated me, so he may fascinate you.

The goal: share a book summary

Some readers of my blog are interested in Daniel P Brown. I wanted an overview into two related Dzogchen classics that had influenced his work:

  • Pith instructions for A Khrid Dzog Chen

  • Self-Arising Threefold Embodiment of Enlightenment of Bon Dzog Chen Meditation

First part of blog post: personal view

I wrote the first part manually, occasionally supported by Deepl to improve my writing as a non-native speaker/writer of English.

Second part of blog post: AI generated summaries

For this part, why should I spend hours and days writing something that AI can do faster and better?

So, I used Google´s NotebookLM to generate the summary. But, to get there, more was necessary. Below, the full toolset that I normally use.

Tools involved




In this case, I bought both books on Amazon. Obviously I can use other sources too (anything)

Used for: manual writing. A translation and writing improvement tool that I have subscribed to (7 euro/month). This is for spell-checking etc.

Used for: audio transcription. An I that I use (when needed) to transcribe audio or video (not used in this case). It uses the AI "Whisper" with its faultless transcriptions . I downloaded the core Whisper AI to my PC to save 20 euro/month but so far turboscribe is just too convenient through its interace.

Used for: writing and text management. A free authoring tool that is amazingly good at organising and extrating text (I use it like a combination of "CMS" - content management tool - and writing tool for books)

Used for: document formatconversions. A free downloadable tool for book management. I use it mainly to convert text between different formats. Instead, one can use free online tools.

Used for: research and summaries. Google free AI for research

used for: blogging. Currently my blogging tool. May not be optimal, I should perhaps go to Medium or so. Too expensive for the use.

For illustrations

The production process

So these were the steps I went through.



Enable NotebookLM (one-off)

It´s free, just chose, but I needed a VPN to make it work in Germany

Download book/sources

This is just one possible source. Also, this is only possible if downloading the book locally is allowed for the book. Most don´t allow it - sadly. This is why I may buy elsewhere in future.

Convert book

I can´t load ePUB into NotebookLM, so it needs to be converted to txt, pdf, docx. For this, I use Calibre

Store books

Store the .docx in Scrivener. I do this so that i can (for example) split the text up into intro and the rest, or into separate chapters as in the original book.

Extract text

Extract all or parts of the book as .txt file for upload to NotebookLM

Upload text

Upload to NotebookLM into a "notebook". This could be a separate notebook for the book, or the author, or a notebook combining several books etc.

Request summaries

Ask for whatever output I want (eg. comparisons, summaries). This may need an iterative approach until I asked the right question at the right level of detail.


as the name sais ...

Proof read

Perhaps already done at step "request summaries", but in generall all AI text should be read to figure out possible errors

Final format

Embellish the format

Add graphics

Using Easil, create a graphic


A thought on...

bottom of page