Archive note (September 2026): First published in September 2024. My NotebookLM and Descript property-report experiment and recording remain, with a checklist for reviewing similar work before sharing it.
As you might have seen on LinkedIn in September 2024, I posted something just a little bit cheeky:
My feeling at the time was: I know CoreLogic releases heaps of valuable information—and we are all grateful for it—but I swear their auction reports have not changed in more than ten years. Is it a case of if it ain’t broke, don’t fix it, or could we maybe check in with the customer every now and then to see how they like it…
Anyway … if you take a look at the comments, you’ll see that the people who watched the video I made thought it was pretty cool.
How did I turn that source material into a video in this quick experiment? I used two AI tools:
Google’s NotebookLM
Descript, one of my favourite tools at the time
The Magic of Google's Notebook LM
In the original experiment, I used Google’s then-new tool to:
Upload multiple sources (PDFs, websites, text)
Generate engaging AI conversations between two “podcast hosts”
Create FAQs, study guides, and timelines
Chat with your content across sources
Back in September 2024, I had only scratched the surface of its potential, but I was already obsessed!
Bringing the audio to life with Descript
Once you have your AI-generated content, how do you turn it into a polished video? That's where Descript comes in. This powerful tool enables you to:
Transcribe your audio
Add visually appealing backgrounds
Incorporate captions and waveforms
Edit and correct on the fly
Create short clips for social media
Those editing steps helped turn the experiment into a video people could watch. The finish still depends on the source material and the checks you make.
A workflow worth repeating
Choose a report you have permission to use and record its publisher, area, reporting period and release date.
Add the source material to your notebook. Ask what is covered and what is missing before generating an overview.
Generate the audio, then listen alongside the report. Check figures, dates, geographic scope and any implied conclusions.
Edit the approved audio in Descript or your chosen editor. Check names, captions, charts and visuals against the source.
Identify the narration as AI-generated, credit the report and make the reporting period obvious. Review the final export before publishing.
Google’s current Audio Overview guidance explains the available controls and warns that generated audio can contain mistakes. A convincing conversation can still misstate a number or leave out an important qualification.
For a property-market update, keep facts and your own interpretation distinct. An AI host is not a witness, a market expert you interviewed or a substitute for the source report.
The Future of Repurposing Content?
As I explored these tools, I had the usual nagging thoughts, like, “Am I out of a job as a podcast host now?” The capabilities of AI in content creation are truly astounding, and they’re only getting better.
The interesting question for me is what a different format helps the audience understand. Start with something worth explaining, then decide whether audio, video or a simple chart does that job best. For another approach, see Creating SOPs with AI.