Using Google's NotebookLM to Produce a Podcast and Summarize Docs | Episode 084

WEBVTT
00:00:22.977 --> 00:00:26.484
Today I've got some over-the-shoulder AI fun for you, so stay tuned.
00:00:26.484 --> 00:00:40.158
Once again, welcome to the Digital Customer Experience Podcast with me, alex Terkovich, so glad you could join us here today and every week as we explore how digital can help enhance the customer and employee experience.
00:00:40.158 --> 00:00:47.477
My goal is to share what my guests and I have learned over the years so that you can get the insights that you need to evolve your own digital programs.
00:00:47.477 --> 00:00:55.012
If you'd like more info, need to get in touch or sign up for the weekly companion newsletter that has additional articles and resources in it.
00:00:55.012 --> 00:00:57.560
Go to digitalcustomersuccesscom.
00:00:57.560 --> 00:00:58.924
For now, let's get started.
00:00:58.924 --> 00:01:13.144
Greetings and welcome to episode 84 of the digital cx podcast, the show where we talk about all things digital in CX, and it's a solo episode 84, mainly because the next two episodes are actually going to be solo.
00:01:13.144 --> 00:01:15.959
So 84, 85 are both going to be solo.
00:01:16.590 --> 00:01:24.986
It is Christmas Eve today if you're listening live, or on day of release, so Merry Christmas to those of you who celebrate.
00:01:24.986 --> 00:01:45.402
Hopefully this is a time for you to spend with family and friends and loved ones and all of that, but I also recognize that a lot of us need a little bit of escape during this time as well, so maybe I am part of your escape today, in which case, thanks for joining me.
00:01:45.402 --> 00:01:47.536
Today is going to be a little bit different.
00:01:47.536 --> 00:02:05.862
And be forewarned if you are a regular podcast listener in other words you're listening to the show, you're not watching it on YouTube you may want to take a second to go find a screen or go look for this episode on YouTube.
00:02:05.862 --> 00:02:11.016
I will be putting a link in the show notes below so you can get to it pretty easily.
00:02:11.016 --> 00:02:37.733
The reason for this is I'm going to do a little bit of screen sharing today and we are going to have some fun with Google's Notebook LM, which, if you haven't used it yet, is a pretty cool Google product, a little bit under the radar, uses Google's Gemini platform to essentially do some summarization of whatever you feed it and also produces.
00:02:37.733 --> 00:02:57.375
This feels really random to me, but it also gives you the ability to produce a co-hosted podcast episode with two a male and a female voice bantering about whatever it is you fed it in terms of the sources, so we're actually going to have some fun with that today.
00:02:57.375 --> 00:03:06.521
In this episode, we are going to feed Notebook LM some sources and do a little bit of prompting, at which point.
00:03:06.521 --> 00:03:13.893
We're going to ask it to produce one of these podcast episodes and on next week's show you're going to hear the result of that.
00:03:13.893 --> 00:03:17.889
So you're going to hear the output of whatever it is we feed it today.
00:03:17.889 --> 00:03:20.881
So you're getting back-to-back solo episodes.
00:03:20.881 --> 00:03:27.960
Today is going to be nuts and bolts and over the shoulder and next week is going to be literally the output of whatever it is we're working on today.
00:03:28.241 --> 00:03:42.391
So, before we jump over into the screen share, I thought it would be good to spend a couple of minutes just talking about Notebook LM and use cases for Notebook LM, because essentially, what it does is takes whatever inputs you give it.
00:03:42.391 --> 00:04:06.712
It could be a website, it could be linked to a YouTube video, it could be a document like a PDF or a text document that you upload, it could be any number of things that you feed it, and it allows you to summarize those documents and essentially have a conversation with Notebook LM about those documents, and then you can pin those conversations so that you can reference them later.
00:04:06.712 --> 00:04:23.295
It's really cool and it sets notebook alum is different than, let's say, chat gpt, for instance, because of this ability to summarize multiple documents into one place and then pinning those things essentially for for later.
00:04:23.295 --> 00:04:30.057
So the obvious use case here is summarizing documents.
00:04:30.057 --> 00:04:36.430
So if you have a swath of documents, you can put those in and summarize them.
00:04:36.430 --> 00:04:41.890
So, for instance, you could well I suppose you could feed it your knowledge base articles or something like that.
00:04:41.890 --> 00:04:55.221
There's a limit, I think it's 20 documents that you can upload to it, but you can point it at a number of websites on the same topic and then have a conversation with it to summarize those things.
00:04:55.603 --> 00:05:06.870
You can also use it to transform documents into something more digestible, maybe turning a book that you wrote into an ebook or a short version of something.
00:05:06.870 --> 00:05:09.556
You can use it to analyze reports.
00:05:09.556 --> 00:05:16.760
So if you have a bunch of data hanging around, you can use it to essentially have a conversation with your data, which is super cool.
00:05:16.760 --> 00:05:25.789
And I mentioned earlier you could summarize knowledge bases, but you could also use it to create a knowledge base.
00:05:25.789 --> 00:05:42.303
So if you have documentation that's pretty broad on a specific niche topic or a group of topics, you could then use it to essentially create outlines or create the startings of different knowledge base articles.
00:05:42.303 --> 00:06:02.197
One interesting use case um might be uh, if you're if you're an average cook, for instance, you could feed it your recipes and have it be a place to store those things and then use it to basically query those recipes if you need some ideas about things that you like doing, so that you know there's just the.
00:06:02.597 --> 00:06:10.918
It's one of those tools where it's so kind of open-ended that really the only limitation is your creativity in actually using it.
00:06:10.918 --> 00:06:15.174
One of the things that Google does suggest that you can use it for is fact-checking.
00:06:15.174 --> 00:06:24.599
I'm on the fence about that, because I don't generally trust generative AI or AI models to do the fact-checking.
00:06:24.599 --> 00:06:31.717
In fact, quite the opposite, I tend to fact-check the outputs of generative AI.
00:06:31.717 --> 00:06:34.483
But hey, you know, go play with it.
00:06:34.483 --> 00:06:42.933
I wanted to at least provide you with a couple of examples of how you might use Notebook LM, either for your personal use or within a team.
00:06:42.933 --> 00:06:47.243
It's an incredibly powerful tool, so, without further delay, let's get into it.
00:06:47.243 --> 00:06:50.397
I'll do a little bit of screen sharing and we'll get into Notebook LM.
00:06:50.397 --> 00:06:59.094
This isn't gonna be necessarily a tutorial like a full-blown tutorial in Notebook LM, but we're just gonna play around and see what we come up with.
00:06:59.971 --> 00:07:02.139
Hey, I wanna have a brief chat with you about this show.
00:07:02.139 --> 00:07:09.097
Did you know that roughly 60% of listeners aren't actually subscribed to the show, on whatever platform they're listening to it on.
00:07:09.097 --> 00:07:14.421
As you know, algorithms love, likes, follows, subscribes, comments, all of that kind of stuff.
00:07:14.421 --> 00:07:25.355
So if you get value out of the content, you listen regularly and you want to help others to discover the content as well, please go ahead and follow the show, leave a comment, leave a review.
00:07:25.355 --> 00:07:32.721
Anything that you wanna do there really helps us to grow organically as a show.
00:07:32.721 --> 00:07:38.793
And while you're at it, go sign up for the companion newsletter that goes out every week at digitalcustomersuccesscom.
00:07:39.349 --> 00:07:40.411
Now back to the show.
00:07:40.411 --> 00:07:47.507
All right, so I've gone to notebooklmgooglecom and sign in with your profile.
00:07:47.507 --> 00:07:48.709
Pretty straightforward.
00:07:48.709 --> 00:08:06.067
I've deleted the ones that I've put in here just to show you what the experience is right off the bat, but it does prompt you to create your first notebook and talks about uploading documents and having conversations with them by having, you know, convert them to useful documentation.
00:08:06.067 --> 00:08:14.056
What it doesn't tell you about here is the fact that it can produce this miraculous like podcast episode that a lot of people are talking about.
00:08:14.056 --> 00:08:19.375
By the way, I can't take full credit for this and I didn't discover this by any means.
00:08:19.375 --> 00:08:37.187
It's pretty well known thing now that you can create these podcast episodes from Notebook LM, but most notably, marley Wagner and Bob London talked about this at Churn Zero and actually produced an episode live during their session, which was super cool.
00:08:37.414 --> 00:08:51.274
Anyway, we'll stop blabbering, I'm going to hit Create and this will pop me into this initial modal where I can start to feed Notebook LM certain sources.
00:08:51.274 --> 00:09:00.129
And you see, here I can upload PDF, txt, markdown, audio, so any number of different things.
00:09:00.129 --> 00:09:04.765
I can upload documents from my Google Drive.
00:09:04.765 --> 00:09:06.102
I can link to websites.
00:09:06.102 --> 00:09:06.586
I can link to websites.
00:09:06.586 --> 00:09:08.336
I can link to YouTube videos.
00:09:08.336 --> 00:09:12.956
Now, what's interesting about the YouTube thing is I was researching this a little bit more you can't.
00:09:12.956 --> 00:09:17.725
It's not crawling video, um, like live.
00:09:17.725 --> 00:09:29.537
It's actually taking the transcripts from those videos, and it actually, um won't take transcripts from, like very recent videos because I guess they need some processing time or whatnot.
00:09:29.537 --> 00:09:37.058
But you are limited to 50 different resources, which is a ton, or you can just copy and paste text in here, right?
00:09:37.058 --> 00:09:40.665
So there's a lot of capability here.
00:09:41.707 --> 00:10:01.984
Now, in thinking about what I wanted to do for this little demo is, you know, there's this concept of wrapped or like end of year kind of recap of what happened throughout the year and I thought, well, why don't I create that?
00:10:01.984 --> 00:10:12.298
But using Notebook LM in audio format, that, but using Notebook LM in audio format.
00:10:12.298 --> 00:10:33.347
And so I think what I'm going to do is feed Notebook LM a series of things that I've done over the course of the year, and we're going to start with just the website, because it will go crawl the website and pull stuff in there as well, stuff in there as well.
00:10:33.347 --> 00:10:38.415
So I'm going to give it digitalcustomersuccesscom to start with and there we go.
00:10:38.415 --> 00:10:46.049
So you'll see, on the left it is looking at digitalcustomersuccesscom and right off the bat in the chat window, so you see, the sources are over here on the left, I can add sources here In the chat window.
00:10:46.049 --> 00:10:54.207
It has started to basically say hey, look, here's kind of a brief summary of the topic, right?
00:10:54.207 --> 00:11:03.299
So in this case it thinks that I'm essentially talking about the podcast, or want to talk about the podcast specifically for this, which it's not wrong.
00:11:03.299 --> 00:11:09.861
It is interesting the title that it gave this chat, which is generative AI for customer success managers.
00:11:09.861 --> 00:11:17.600
It probably did that because I have my AI tutorial for CSMs embedded on the homepage of the website.
00:11:17.600 --> 00:11:24.398
Now that begs the question should I maybe feed it some sub pages?
00:11:24.398 --> 00:13:11.490
So let's go look at digitalcustomersuccesscom and what I might do is point it to the four pillars of digital customer success, which is an article that's on there.
00:13:11.490 --> 00:13:19.173
I'll go ahead and add that as a source as well, just to see how reactive this thing is.
00:13:19.173 --> 00:13:26.814
I'm still kind of in play mode here, if you couldn't tell, but we're playing live together, super fun, all right.
00:13:26.814 --> 00:13:27.836
So I fed it that.
00:13:28.316 --> 00:14:00.540
Now, in thinking about what other content I wanted to feed this chat you know, to give it some sense, for you know the, the show and kind of what we've been through on this show journey together I thought it was good to feed it a few episodes, and so what I will do next is essentially give it some links to some of my solo episodes that I've done, and so I will start adding those as sources here.
00:14:00.540 --> 00:14:17.287
Now I'm not going to drag you along while I add, like I don't know, 10 different episodes here, but, as you see, here I've added a YouTube video and you can check and uncheck these as you go along, right.
00:14:17.287 --> 00:14:26.133
So what I'm going to do real quick is just take a second and go add probably half a dozen or so solo episodes.
00:14:26.133 --> 00:14:40.855
The other thing that I thought I would do is add a list of guests that I've had over the years Sorry, not over the years, over this past year and see if that helps the cause as well.
00:14:40.855 --> 00:14:41.686
So give me a second.
00:14:41.686 --> 00:14:45.394
I'm going to do a bunch of work and then I'll see you here shortly.
00:14:45.394 --> 00:14:49.692
All right, so I've added a bunch of YouTube episodes.
00:14:50.234 --> 00:14:54.105
As you can see, I've done a couple of other things that I'm going to add here.
00:14:54.105 --> 00:15:02.300
First of all, I am going to upload a list of guests that I've had.
00:15:02.300 --> 00:15:07.338
I just put that into Google Drive guests in 2024.
00:15:07.338 --> 00:15:09.752
I'm going to insert that.
00:15:09.752 --> 00:15:19.620
So I think that gives me a general yeah, it gives me a general kind of overview of the show of this year.
00:15:19.620 --> 00:15:34.740
It's got quite a few episodes, it's got the web page, it's got my guest list for the year, and so now I can start to interact with these documents and these videos and start typing and have a conversation with it.
00:15:34.825 --> 00:15:38.373
Now, before I do that, I did want to show you real quick.
00:15:38.373 --> 00:15:47.523
The output element of this is on the right-hand side here, and so you know you can produce various things.
00:15:47.523 --> 00:16:09.157
For instance, I could produce an FAQ if I wanted to, which is super powerful, you know a briefing document, if I wanted to present an executive overview, for instance, of the podcast to somebody, or a study guide, if I was going to offer a certification on all the topics that have been discussed over the last 84 episodes, which I don't even know if I could pass.
00:16:09.157 --> 00:16:10.707
But that's beside the point.
00:16:10.707 --> 00:16:14.235
Anyway, there's a bunch of outputs here.
00:16:14.235 --> 00:16:24.060
One of them is this deep conversation so two hosts, english only, and that is essentially what we're going to output for next week's show.
00:16:24.060 --> 00:16:26.954
I'll play a clip of it here when we're done.
00:16:27.644 --> 00:16:39.081
But the first thing we're going to do is just have a conversation with Notebook LM so that we can start to direct what we want that podcast episode to talk about.
00:16:39.081 --> 00:16:41.947
And so it does give you a couple of suggestions down here.
00:16:41.947 --> 00:16:44.315
How does generative AI impact customer success?
00:16:44.315 --> 00:16:50.418
How can community building enhance digital customer success, all that kind of stuff.
00:16:50.418 --> 00:16:54.729
This is all digital customer success related, which is great, because that's the subject matter.
00:16:54.729 --> 00:17:08.206
However, right now, what I want to do is basically ask it for an overview of what the Digital CX podcast is, and so we're going to ask it.
00:17:08.206 --> 00:17:24.000
Can you give me a summary what the Digital CX podcast is and the major themes covered on the show.
00:17:24.000 --> 00:17:25.882
I'm going to see what that comes back with.
00:17:25.882 --> 00:17:39.916
All right, now, what's cool here is it has come back with a pretty lengthy list here, but it does tell me about the podcast itself, and you'll notice that it does cite sources, which is incredibly cool.
00:17:39.916 --> 00:17:43.689
Think of this as basically like a steady guide.
00:17:43.689 --> 00:18:00.100
So if you fed this a bunch of source documentation and asked it to summarize it, it would do the same thing that it's doing here, which is it'll give you the sources from where those things came.
00:18:00.100 --> 00:18:02.672
So very, very cool.
00:18:02.672 --> 00:18:12.460
And then, yeah, as per my prompt, it does then tell me you know what are some of the major themes that the show talks about.
00:18:12.904 --> 00:18:19.345
And now what I'm going to ask you to do is to give me a recap of 2024.
00:18:19.345 --> 00:18:23.689
Is to give me a recap of 2024.
00:18:23.689 --> 00:18:26.351
And this is where I'm going to manually actually feed it.
00:18:26.351 --> 00:18:32.017
My Spotify wrapped details, so we're going to pepper in a little bit of manual information here as well.
00:18:32.017 --> 00:18:45.510
You can see, the show grew phenomenally, by the way, which is awesome 316% follower growth, which is awesome, awesome.
00:18:45.510 --> 00:18:51.663
My average five star rating is five stars, so thank you to those of you who've left a five star review.
00:18:51.663 --> 00:19:23.096
Um, all right, so I've skipped ahead a little bit, because it does take it a second to really come up with this stuff, but you'll see that it says we've experienced significant growth, which is all accurate Increase in the show's relevancy and impact and digital customer success community, as well as a commitment to providing valuable insights to its audience.
00:19:23.096 --> 00:19:31.038
Awesome recap of the journey, name change and scope of expansion uh, yeah, this is great.
00:19:31.038 --> 00:19:41.591
So this picked up from episode 50 that I fed it because that's when we changed the name from digital CS podcast to CX podcast, so it's great.
00:19:41.591 --> 00:19:43.676
Uh, focus on the four pillars.
00:19:43.676 --> 00:20:10.065
We did a lot of stuff on the four pillars this last year emphasis on measuring, roi, exploration of automation, generative AI integration which is cool community building okay, so I mean a decent summary.
00:20:10.125 --> 00:20:24.317
Now, what isn't apparent here is that you know this this isn't going and crawling the web for new sources, right, it is specifically looking at the sources that I am feeding it.
00:20:24.317 --> 00:20:27.942
So could it perhaps produce a better result if I fed it some additional things?
00:20:27.942 --> 00:20:37.084
Absolutely, did I think about the possibility of feeding it every single transcript from this year's episodes?
00:20:37.084 --> 00:20:41.771
Single transcript from this year's episodes?
00:20:41.771 --> 00:20:43.013
Absolutely, and we probably could have done some of that.
00:20:43.013 --> 00:20:53.413
But also, you know, using tools like this, the whole goal is to create some efficiency in your life.
00:20:53.413 --> 00:21:09.756
If I were to go and download every single show transcript, if I were to go and download every single show transcript format it put it in here you could argue that I could really just as easily in that time record a recap episode of myself.
00:21:09.756 --> 00:21:12.041
So you know, there's a point of diminishing returns here.
00:21:12.041 --> 00:21:22.389
But the point is that you can feed it a lot of information and have it distill that information down into digestible chunks and really good outline format.
00:21:22.549 --> 00:21:42.794
Now, one of the features of Notebook LM here is that it allows you to save notes, and so I'm going to click save note and what that does, because as you have a conversation with the chatbot, you know things get kind of lost in the mix, and a lot of times with ChatGPT you have to go back and reference things that you did.
00:21:42.794 --> 00:21:52.424
And one thing that makes this really a notebook is that you can save these notes over here for reference later.
00:21:52.424 --> 00:21:53.826
So super, super handy.
00:21:53.826 --> 00:22:01.332
So I'm going to leave it at that for now.
00:22:01.332 --> 00:22:01.513
I could.
00:22:01.513 --> 00:22:09.457
I could go on and refine things, and it's it's best practice that you do that with any uh, you know, with any ai, but I feel relatively good about where we stand with our results thus far.
00:22:09.697 --> 00:22:19.853
So what I'm most interested in here today is this audio overview deep dive conversation to hosts english only I would imagine they're going to probably going to probably going to expand that at some point.
00:22:19.853 --> 00:22:26.814
But what I'm really interested in is having this thing create next week's episode for me so that I don't have to.
00:22:26.814 --> 00:22:51.887
So we're going to have two hosts talk about the Digital CX podcast a year in review, and so I'm going to hit the customize button because then so I'm going to hit the customize button because then this will give me the ability to basically prompt what I want these two hosts to talk about in this episode.
00:22:51.887 --> 00:22:52.269
And so here we go.
00:22:52.269 --> 00:23:09.956
Be a recap, let's say, of the year 2024 as it pertains to the Digital CX podcast.
00:23:09.956 --> 00:23:29.096
The episode should highlight some of the major milestones and accomplishments achieved throughout the year, as well as highlight the concepts.
00:23:29.096 --> 00:23:42.751
I can't spell concepts discussed throughout the year High level, but we're going to give that a try.
00:23:42.751 --> 00:23:44.855
I'm going to hit generate and it's going to chug on that for a little bit.
00:23:44.855 --> 00:23:50.293
So while it's doing that, I hope you see how cool this tool is.
00:23:50.293 --> 00:23:58.534
I really want to use it a lot over this coming year as a notebook and helping me summarize certain things.
00:23:58.534 --> 00:24:10.801
When you exit out of a specific one that you're working in, you'll see that it has a placeholder for this particular conversation that you're having with Notebook LM and then you can add to those right.
00:24:10.801 --> 00:24:26.721
So it's not this stream of consciousness, there's this concept of just separate conversations that you're having within Notebook LM, but I think you're going to be pretty impressed with what it pumps out as its podcast episode.
00:24:27.903 --> 00:24:30.673
I found this to be fascinating for a couple of reasons.
00:24:30.673 --> 00:24:38.000
First off, it just seems random, you know, to have it pump out a podcast episode of all things.
00:24:38.000 --> 00:24:43.356
You know the study guide makes sense, the briefing doc makes sense, faqs that makes perfect sense.
00:24:43.356 --> 00:24:46.182
But then you know it's a podcast episode.
00:24:46.182 --> 00:24:47.897
So somebody at Google, I think, was having fun with this.
00:24:47.897 --> 00:25:02.805
On the surface it's an incredibly cool thing and for the most part the output is relatively engaging, though when you really listen to it it can be a bit of snooze fest.
00:25:02.805 --> 00:25:13.006
To be honest with you, the hosts are engaging and they sound like podcast hosts and it sounds really well produced and those kinds of things, but the banter is a little bit contrived.
00:25:13.006 --> 00:25:16.876
You'll get what I mean when you hear the results.
00:25:16.876 --> 00:25:25.017
So I really hope you've enjoyed this brief, over-the-shoulder experience of Notebook LM.
00:25:25.017 --> 00:25:37.474
Maybe you can start to utilize it yourself as well, and what we're going to do is actually outro this episode with a sneak peek of what it is generating as we speak.
00:25:37.474 --> 00:25:42.993
But, that said, I want to wish everybody a very Merry Christmas.
00:25:42.993 --> 00:25:52.365
If you celebrate, if you don't, hope you get some good time off with the family and all that good stuff and make sure that you tune in next week on Tuesday.
00:25:52.365 --> 00:25:54.530
I guess that would be New Year's Eve.
00:25:54.530 --> 00:26:00.860
Yes, it would be on New Year's Eve, because we're going to be playing the output of Notebook LM in its entirety.
00:26:00.860 --> 00:26:02.242
So hope you enjoy that.
00:26:02.242 --> 00:26:06.836
Thanks for listening today and we'll talk to you again in the next episode.
00:26:07.519 --> 00:26:09.243
This show was busy in 2024.
00:26:09.243 --> 00:26:12.972
We're giving you the VIP backstage pass to their year.
00:26:12.972 --> 00:26:15.880
Yeah, the podcast itself really evolved this year.
00:26:15.880 --> 00:26:20.649
It went from just digital customer success to a bigger picture of digital customer experience.
00:26:20.649 --> 00:26:24.040
Thank you for joining me for this episode of the Digital CX Podcast.
00:26:24.040 --> 00:26:29.209
If you like what we're doing, consider leaving us a review on your podcast platform of choice.
00:26:29.209 --> 00:26:31.678
If you're watching on YouTube, leave a comment down below.
00:26:31.678 --> 00:26:42.678
It really helps us to grow and provide value to a broader audience and get more information about the show and some of the other things that we're doing at digitalcustomersuccesscom.
00:26:42.678 --> 00:26:43.640
I'm Alex Tergovich.
00:26:43.640 --> 00:26:44.750
Thanks so much for listening.
00:26:44.750 --> 00:26:45.695
We'll talk to you next week.