Posted by Apple Podcasts ¦ Amazon Music ¦ Spotify
Hi, and welcome to Try AI for Growth, a baby podcast out of Make Space for Growth. I’m your host, Sara Vicente Barreto, and here I share short and sometimes surprising stories of how I’ve used technology to tackle everyday challenges—at home, at work, and in business
Today’s episode is about a tool that I’ve been curious about for a while—Notebook LLM. I had heard students were using it to summarise notes or turn content into learning aids, but I wanted to test something a little different. What if I asked it to turn one of our NGO reports into a podcast? So many people prefer the audio format, and I wondered if I could do that, without producing it myself, like in this podcast. What I did not know is that it would generate a conversational podcast that was so real that my entire team in Mozambique thought I had done it myself, even without my own voice. Let me tell you what happened.
The base: an annual report
Each year, our HIV Day Centre in Mozambique prepares a detailed annual report—full of data, qualitative reflections, and program updates. These reports are important, but not always the easiest thing to share more widely. They are pretty long, and we tend to share them with the specific donours that have contributed to the project, but not always with the wider community. This year, we had added a short slides version to go along and perhaps share on our social, but I was curious about a new format.
So, I decided to try something new. I uploaded the HIV Centre report to Notebook LLM. My goal was to see what kind of audio podcast it could generate from a document that was never written for audio—or even public storytelling.
My first attempt: Learning the interface
As most of the interfaces, the tool is very user-friendly. But as with all of them, there are always things that are good to know before starting to use it. I found out that once you generate the first version of the podcast, you can’t go back and make changes or edit; you have to just create a new project and upload the documents all over again.
Still, the output? Not bad at all. It took the key content, selected a few themes, and drafted a conversational-style podcast reflecting on the year. It didn’t sound robotic. It sounded relaxed, authentic, and oddly… like something I might have actually said.
I also did not realize I could not chose the output language and, even though the report was fully in Portuguese, given my google settings were in English, it generated an English version of the podcast. The only good part about that is that I can share the link here so you can see the result, though this is not my final product!
By the way, when I go into Notebook LLM today, I can see an interactive version in Beta mode, so I am guessing some of these limitations are changing.
Attempt #2: Giving it Direction
For my second try, I went to the instructions box. I selected the desired duration, designed a prompt (which sadly I did not save, so I still need to work out how to get a similar one for the next report). In the prompt, I gave a few pointers that I wanted it to focus on what the Centre does, what happened during the year and what the impact of this project was.
The outcome? Even better.
I had a thoughtful, conversational recap of the year. It included reflections on successes, challenges, and even mentioned the community impact of specific programs. It had hesitations and surprised reactions. All without me writing or recording a single sentence. I was blown away.
What did people think?
With this, I downloaded the audio and shared it with the HQ team. They could not believe it had been generated while I had gone out to get a coffee. Or less. We all gathered and played the audio out loud in the office. We were all blown away. Other than the word stigma, which it would not sound right, everything else sounded absolutely real.
It was time to go a step further. I shared it with the Mozambique team and our Sister partners at the project. Their reactions were so positive! And when I told them it was AI-generated, they could barely believe it. They ALL assumed I had recorded the episode myself. That, to me, was the biggest vote of confidence. It was clear. Empathetic. Grounded in facts. And it sounded human.
And here’s the thing—I hadn’t touched the script. Imagine if you got to edit and go further.
What do to with it?
You may wonder if there is a real-world application in this. For charity, there are endless options, but for corporations too. An annual report is not always an easy thing to consume. But once you turn an annual report into a conversation, our ability to follow and understand (and relate) to the company is much greater. And it is not just annual reports, you can talk about new product launch briefings, technical reports or even industry reports that matter to you and that we can now relate to the company for our clients or share internally in a more consumable format for training.
For charity, it provides an opportunity to launch a deep-dive series with all the reports for the year across all projects, an opportunity to share on socials as a different format from what we have ever done. And with no technical expertise or expensive production times.
By the way
For reference, I also decided to use my friend ChatGPT to generate a podcast for me. I was curious about the differences and also about our potential preference to record the podcast in our own voice. So I tried 2 additional requests:
First, I provided the same report to ChatGPT and asked for a 5-10 minute podcast script in narrator format to kick off a series of podcasts about our projects. I asked for an intro about the project, the problems we are trying to address, the activities developed, the impact measurement and the outlook for 2025. I got an easy-to-digest short script which I could record just as fast as I do with this podcast. The added benefit is that I could provide input back and edit the final version before recording it (more to come on how to transform the script into a professional voice recording in a future podcast!)
The other thing I tried was to provide the set of slides that summarised the report and could be shared as a carousel on social, and I requested a voice-over for the slides. I found it a bit short, but it was still a helpful start. As an added bonus, I asked ChatGPT to teach me what the best technical solutions would be to execute on this. All in one place.
Bonus: FAQs, Mind maps and more
But let’s go back to Notebook LLM. After exploring the podcast function, I noticed a FAQs button. I clicked it just to see what it would do. Within seconds, Notebook LLM had generated a short FAQ section based on the uploaded report. It covered questions like:
- What does the HIV Centre do?
- Who are the beneficiaries?
- What are the goals?
- Which activities are provided?
- What impact did it have in 2024?
- What were the main challenges?
- How is the centre financed?
We can now think of having an easy project page on our website, addressing FAQs or again publishing them on socials. Just like that—I had another piece of usable, digestible content, and I hadn’t lifted more than a few clicks. If you are less fussed about publishing, you can also just query the model. When uploading files, you can then ask questions about their content directly to Notebook LLM and limit it to only those documents for the answer (rather than the broader world that a ChatGPT would access).
As I was preparing this podcast, I also decided to give mind maps a go, and was really impressed by the result and how it can transform something so dense into clear schematic pieces of information!
Lessons Learned
Here’s what stood out to me after this first experiment with Notebook LLM:
✅ Use the instructions box — It’s your chance to define style, tone, and focus. Don’t skip it.
✅ Set the output language — You need to do this before generating the podcast or FAQs.
✅ You can test without risk — Nothing is final. You can generate, revise, and test again. If you want to try more rounds, you need to delete the previously generated audio and try again
What you can try
If you’re curious about Notebook LLM or want to use AI to amplify your impact, here’s what you can try:
1️⃣ Upload a report, article, or blog post and ask for an audio version in a conversational tone.
2️⃣ Test the FAQs generator for reports, websites, or onboarding packs.
3️⃣ Create a learning recap for donors or board members from longer documents.
4️⃣ Translate internal docs into accessible formats—podcast, newsletter, or even a social post.
5️⃣ Use it as a first draft assistant when you’re stuck or out of time.
I am leaving instructions in the show notes to link into the Google tutorial so you can get a shortcut before starting.
This experience reminds us that AI doesn’t just save time—it creates new ways to share stories. I didn’t need to script a podcast or draft an FAQ page. I just needed a good document and a few simple instructions. So maybe I like to draft and record podcasts, but not everyone does. Maybe they are great report writers but their message is getting lost when it gets out there.
So if you’ve been sitting on a great piece of content but haven’t found the time—or energy—to turn it into something shareable, give Notebook LLM a try. You may be surprised by what it sounds like.
Thanks for tuning into this episode of Try AI for Growth. Don’t forget to subscribe, leave a review, or share how you’re using AI to lighten your workload and expand your reach. Until next time—keep experimenting, keep growing, and keep having fun!