Most AI tools let you customize your user experience to an extent, and NotebookLM is no different. You can guide the way it speaks to you through prompts, but there’s a better way to do it. Earlier this year, NotebookLM got a Learning Guide feature which makes studying more interactive and forces you to think more about what you know. And along with this feature came the Custom mode - this is what enabled this entire experiment.

My colleagues have had success with using NotebookLM as a “TikTok brainrot” tutor through Custom mode, where NotebookLM basically answers back in Gen Z slang to make the process more engaging and enjoyable. This made me wonder how far you can really push this custom mode, and whether I could instruct NotebookLM to communicate with me the way I communicate with it.

Why would I even want to turn NotebookLM into an AI version of myself?

There are some actual benefits to this

Setting up NotebookLM to respond like me is a two-parter: writing style and thinking patterns. Like probably most people in my age bracket, I don’t use the same writing style professionally as I do in my day-to-day life. Getting a message or post out quickly and conveniently matters more than being grammatically accurate. I might regret admitting this, but a couple of examples include not using capital letters and shortening my words where possible.

Then there are thinking patterns and structures. As someone who’s been diagnosed with ADHD, my head isn’t exactly a calm and linear environment, but I’ve got a couple of tools to help me navigate the mess. For example, short bursts of information; if a concept can be summed up in just two words, I don’t want it to be any longer than that. Other patterns include mixing casual shorthand with proper terms, looping back to earlier points, and most importantly, bottom-up thinking.

There are a multitude of benefits to setting up NotebookLM to talk in your tone and thinking patterns:

• It’s familiar - When an AI responds in your natural tone, you don’t get that jarring shift where a tool feels rather corporate.
• Keeping you grounded with context switching - It helps you understand concepts in your own language, regardless of how they’re typically communicated.
• It keeps you honest - When an AI mirrors not just your writing style but also your thought patterns, it can give you a clearer read on your own shortcomings without being judgmental.
• You don’t spiral into perfectionism - If my AI research assistant can explain coursework in the same messy language I use every day, it removes the pressure to perform and sound smart.

Customizing the way NotebookLM talks to me

Instructing NotebookLM is dead simple

If you want to try this out yourself, I recommend creating a new notebook or finding an existing one with only a couple of sources so you can give it a spin. There’s a little settings icon in the top right of the chat panel, which will bring up the configure window set to the defaults - click Custom. And for the length, I selected Shorter. Another nice-to-have is the resize handle; dragging it down will expand the window so you’ve got more visible text real estate. And then you just put in your instructions - I put mine in a list format, and included some official terminology too.

How I’m actually using my AI self

Putting my brain clone to use

I started by trying this out with a couple of tutorials on creating databases in LibreOffice. I asked NotebookLM to explain the steps to me, and it stepped up with the structure and tone I asked for. So I hopped over to my anxiety notebook, configured the Custom mode, and asked it to create a short course. I didn’t like part of the ChatGPT-style answer it gave me: “the most important part of getting started isn't learning a coping skill—it’s changing your goal”. I like a good contrast, but the AI contrastive focus construction is getting boring. Back to the drawing board…

I instructed NotebookLM to avoid the cringy AI-style contrast in its sentences and redid my prompt for the anxiety course. It finally gave me exactly what I’ve been looking for. It explained its choice in four short bullets and in my own context - shorthand with proper terms and looping back into previous points. Plus, it gave me a super easy-to-follow course for “mindset and purpose”, in the bottom-up style I instructed. Now, all I had to do was copy-paste these Custom mode instructions into my other notebooks whenever I wanted NotebookLM to take on my tone and thinking style.

I was surprised at how well my Custom mode instructions held up, even when NotebookLM explained some more technical topics, like implementing mental models in UI design. It only took a couple of retweaks to get it perfect. If you’re trying Custom mode yourself, definitely keep tweaking it until you get the results you want - you can type up to ten thousand characters!

Making NotebookLM mirror my own communication style

The goal of this wasn’t to literally build an AI version of myself, of course, but to attune NotebookLM to the way I think and speak. And I almost can’t believe how well it worked. Most AI customizations don’t feel like they actually change anything. But with NotebookLM, it truly feels like having concepts explained to you on your own wavelength. I usually say "it's all in the prompt", but this time it's all in the Custom mode instructions.