There are few better feelings than a group chat with your friends blowing up randomly, each person adding joke after joke, and you can't help but laugh at the chaos. Over the years, I've inevitably had a bunch of these chats.
While I do remember snippets of each of these chaotic conversations vaguely, I've long forgotten most of the context. That's when I thought: what if I could bring these chats back to life... in a podcast form? I didn't even need to think twice about the tool I was going to use to achieve it -- NotebookLM was the obvious choice.
Here's why and how I set it up, step by step
Though there are quite a few tools now that can convert files you upload into full-fledged podcasts, NotebookLM's Audio Overviews are the most impressive I've tried so far. NotebookLM's specialty is that it's designed to be source-grounded, which means that it only references information you've fed it to answer your queries or generate outputs (including Audio Overviews).
The reason this is so important is that a major problem with the majority of AI tools nowadays is hallucinations, which are essentially when AI confidently makes up details to fill in gaps. You'll notice AI models will often fabricate information just to please you, invent random events that never happened, and link things together in a way that sounds right (even though it isn't).
When it comes to this use case, where I want to convert real conversations I've had into a podcast, I want it to reflect stuff that actually happened instead of made-up narrative. So, this source-grounded element is absolutely crucial and keeps the whole thing honest. Even when the final Audio Overview leans into humor or dramatizes certain moments, it's still pulling directly from your actual messages.
Fortunately, setting this experiment up was as simple as it could be. The only source I really needed for this was the raw chat export itself. WhatsApp is my primary messaging app, so I simply located the group chat I wanted to use for the experiment and used its built-in Export Message option to generate a .txt file.
While there are Python scripts designed to clean WhatsApp chats and make them easier to parse that I found on Reddit (and which is how I came up with this idea in the first place), I didn't end up needing any of them. I just uploaded the messy, unedited export as-is and had no problems.
So, no complex setup required at all! All you need to do then is hit the Audio Overview button in the Studio panel, and wait for NotebookLM to work its magic. Since NorebookLM does let you customize your Audio Overviews by adding a prompt, I decided to explain what exactly I wanted it to generate by adding the following prompt:
I loved the podcast NotebookLM generated
I'm a big fan of NotebookLM's Audio Overviews and generate a bunch daily. Even then, I can confidently say that this is among the best ones it has ever made for me. The group chat data I added had over a year's worth of messages, and the AI tool managed to pull out both the most important moments and the funniest little details without getting lost.
For instance, I've unfortunately missed a bunch of gossip sessions that went down in the chat over the past few months due to my hectic schedule. When I get the chance to open them, the chat is filled with hundreds of thousands of messages, and I usually just give up and scroll to the bottom.
Funnily enough, I found out about some of the gossip for the first time through the Audio Overview itself. It pointed out a few wild moments I somehow completely missed, and hearing them narrated back to me in a podcast-style format made it even funnier.
The friends in my group chat are all college students like me, which means a lot of our conversations involve college. The hosts pointed that out and highlighted how the shared college trauma bonded us all. One thing I loved is how the hosts managed to pull out relevant examples for each of the points they made.
For instance, in connection to the college bit I just mentioned, the hosts mentioned an example of one of my friends asking for prayers right before a dreaded math quiz. Similarly, they also pointed out that I had voiced in the group that college exams are created in a way that is mathematically impossible to complete in the allotted time, setting you up for failure! What's even funnier is that I don't even remember saying that at all! Hearing the hosts quote it back so dramatically had me crying laughing.
They also brought up my absolute disaster of a day when I missed my bus, ended up sprinting like my life depended on it, and then immediately spilled coffee all over my hoodie. The hosts even reminded me that my mom had told me to just take the day off, but of course, I decided to run anyway. Hearing that whole sequence narrated back was hilarious. All of this also made me realize that I'm the most dramatic person in the group chat!
I found this new NotebookLM feature so good, I might stop using all my other productivity apps
Might be time to make NotebookLM my only productivity tool.
Posts 2
By Mahnoor Faisal
Sep 27, 2025
A pretty significant moment in my friend group was one of my friends moving to Malaysia. That was obviously discussed a lot, and the hosts managed to give me a fun rundown of everything -- from her first announcing it, the farewell party we threw for her, her last day before moving, and even her post-move messages once she landed. It's been almost a year since she moved, and I felt as emotional as I did the day she left.
Of course, there's a lot I can't mention in this article since the chats consisted of private jokes, inside references, and some really embarrassing moments that only our group would understand. Even so, the Audio Overview did a fantastic job of capturing the essence of all these memories without ever needing me to explain the context.
I also generated a Slide Deck (which is a newly launched NotebookLM feature) of the chat, and that was another incredibly fun addition.
The visuals were on-point, and I had so much fun going through it. It gave different nicknames to each member of my friend group, broke down different gossip sessions we've had over the months, and even pulled out relevant messages for each category it created. It honestly felt like a personalized year-in-review of our group chat, and I couldn't stop smiling while flipping through it.
I'm so glad I did this
It's been a couple of months since my friends and I last got together, and this was a great way to relive all the chaos, laughs, and little moments we usually miss in our day-to-day lives. The only con really was that it made me miss them even more than I already did.
