AI’s Impact on Publishing

Guy: [: 00:00:00

I am live today and so is Keith. We are not agents, we are not ai, but we're actually gonna be talking, but it's kind of is a joke of what's uh, could be possible in the future. In terms of how to, uh, use AI as a, potentially as an agent or just as a function to be able to take over and do various things.

ooks, and also Perfect Bound [: 00:01:00

But on the other hand, I think there's, I think both he and I think there's a lot of opportunity on there. So let me tell you a little bit about Keith. He is the CEO of the Brooklyn based independent publishing companies. As I mentioned, Ulysses Press and Velo Press books. He's also the CEO of Perfect Bound Technologies Corporation and co-founder of the book Publicity and Digital Marketing firm, Pacific and Court.

. Keith, welcome. Uh, what a [: 00:02:00

Looking forward to, uh, your perspectives on, uh, marketing and ai. Thank you so much. I'm really happy to be here. Yeah. Thank you. Uh, so, uh, tell us how did you get involved in, uh, marketing and ai? What's your backstory?

Keith: Uh, so my backstory is my parents founded Ulysses Press 40. One years ago now, uh, my dad did everything he could to push me away from the publishing industry.

ublishing company starting in: 2001

Did a short stint at [00:03:00] Ette Book Group as their associate director of analytics. So that's how I really got into marketing. Uh, being at a Big Five company and being able to oversee a lot of the marketing strategy that was happening at at ette allowed me to understand, you know, what was working and what wasn't in.

Book marketing, and then I brought that back to Ulysses when I returned as publisher and now CEO. You know, marketing is a huge challenge in book publishing, and it is one of our core focuses, especially as a data-driven and trend-driven publishing house.

Guy: Absolutely. And I'm so glad to hear that you're combining marketing and analytics in the same sentence.

, and some of the other, uh, [: 00:04:00

Keith: books?

So Ulysses Press is a general trade nonfiction publisher. So we do a lot of work in pop culture, cookbooks, uh, health and fitness. And then Velo Press is the premier publisher of hardcore exercising books, basically. So triathlons, ultra marathon running, uh, cycling, that is the, the core of fellow press. And when I'm.

Looking for new projects, or in the case of Velo Press, which we acquired from Outside Magazine, I'm always looking for publishers that have category killer projects. So the ultimate book on how to train for a triathlon or the ultimate Italian cookbook. 'cause the, the biggest challenge in publishing is, you know, if you.

Look at it from a brand [: 00:05:00

So if you're not able to establish that brand in the marketplace, the book really goes nowhere. And that is by far the biggest marketing challenge with. Publishing.

Guy: Yeah. Interesting. And, uh, it makes a lot of sense where then you get the synergies and, and to your point, you know, when you're trying to market a book and I've marketed now four or five books and it is hard, you know, just as a, as an independent, um, author, it is really hard.

nd you're exactly, you know, [: 00:06:00

Keith: I think that it's going to, uh, begin to affect. Every backend operation of publishing, I don't think that there is any chance in the near future that you're gonna have, uh, professional publishing companies. Uh, putting AI generated content out into the market, even as we move into an era of copyright laundering where one, you know, AI, LLM is teaching a smaller LLM, what it knows and you're able to get away from the original copyrighted material.

n of that, and that is gonna [: 00:07:00

Amazon a couple of months ago released Rufuss, which is their AI chat bot. On Amazon where you can ask a question of it. You know, I'm looking for a book for my brother-in-law. He's really into phishing, and instead of giving you this endless scroll of results, it's gonna give you three of its best recommendations and that.

Has the potential to really transform the way that we market books and the way that we position them online. Considering Amazon is 50 to 60% of most publishers, uh, annual revenue.

. Is there then a, a way to, [: 00:08:00

Keith: I think that it, it's already happening. And by the end of SEO, I'm, I'm really mean the end of SEO as we've known it for the last 15 years or so, uh, which has enabled a lot of publishers to foster awareness of their books without having to spend money. We are already seeing on Google and Amazon and other search engines, the.

ishing company that may have.: 2000

The other thing that I really think is that the. Incentives are not quite aligned for book publishers and companies like Amazon. You know, book publishers want to sell through all of their SKUs, whereas Amazon, realistically, they wanna limit the number of SKUs that are in their warehouse. You know, their logistics operation is what a $40 billion cost to Amazon, so the better that.

They can get at delivering the exact product to the consumer in as easy a fashion as possible for them. Uh, the more they stand to make, and unfortunately that is going to really limit the ability to discover products on platforms like Amazon.

Guy: Yeah. But how does that, uh, work when, um, I mean the cost of carrying.

t high. When you think about [: 00:10:00

Keith: I totally agree with you. Um, I think that. KDP is going to become a very important driver of book sales going forward for exactly that reason. The issue with publishing is most of the books that we are selling are printed offset, you know, in China or in the US or Canada and shipped to the warehouse and, and held there.

Uh, so that is the cohort of books that I think are gonna come under the biggest threat. Yeah, from, from

y said, well, we're gonna do [: 00:11:00

Yes. Uh,

Keith: yeah, which I think is gonna be a big use case for ai. Uh mm-hmm. Already in, in our organization, just using, uh, tools like copilot or Chat, GPT forecasting, demand for books is notoriously difficult. We use a, uh, multiple regression equation on all of our books. We take this, mm-hmm. Very large data set every year to see how our first year sales of books compared to the estimates that we got and the number of books that we printed.

can optimize it, I think the [: 00:12:00

End up pulping your excess inventory every year, and it's a huge, uh, cost.

Guy: Yeah, yeah. Well, I see, uh, publishers kind of as a, uh, venture capitalist, you know, when you think about venture capital, they invest in 10 companies to expect one, you know, to really make it two or three to do Okay. And then six or seven to kind of, you know, fall by the wayside.

And, and I don't know how the numbers are with you, you got, I imagine it's.

Keith: So the best publishers in the world bat 300, you know, so three outta 10 books make it and pay for the other seven failures. Uh, most publishers are shooting to bat 200, uh, which is it? It's pretty crazy. Yeah. You're investing $50,000 or a hundred thousand dollars up for.

e just never gonna earn out. [: 00:13:00

Four or $5, and that four or $5 has to pay for salaries, rent, you know, operations, insurance, as well as marketing. So you're talking about having to build a unique brand for, you know, one title that may or may not work. And you have about 50 cents per copy to spend on that marketing before you start losing money.

s fail and is one of the big [: 00:14:00

D two C sales through, through your website, optimize for the books that are working and ignore the books that that aren't. Uh, but that is the, the big challenge in book publishing.

Guy: Uh, when I think about marketing a, a brand, so a book as a brand or an author as a brand, then you know, a lot of the marketing that you're talking about, you know, whether it's SEO or, or now maybe with Rufuss, that isn't gonna get you more than.

f hundred percent, you know, [: 00:15:00

Keith: I. I mean, right now it's TikTok, you know, talking about ai. Uh, that is the, the biggest driver of book sales. Uh, we are in a very weird moment for marketing and publishing. Used to be that a New York Times ad or a New York Times book review, uh, getting your author on the Today Show. That would just blow up a book.

n it went right back to the, [: 00:16:00

So that is the, the biggest challenge in publishing right now is being able to figure out what does work and how you can leverage that to significantly increase your. Discoverability, uh, Amazon continues to be a, a core driver of that using Amazon advertising. Uh, again, a great area for the future of AI is being able to really actively and accurately micro target the people who might be able to enjoy your book.

But it's just, you know, it's a game of whack-a-mole. It's really throwing darts in the dark when it comes to. To marketing and, you know, doing it across so many SKUs and having lean teams who are doing that, it's a, it's a really good use case for AI to be able to automate a lot of the analysis of what is working across all of these different marketing strategies.

You know, [: 00:17:00

Guy: Yeah. Yeah. Now you mentioned, uh, TikTok. How is AI or how do you see AI helping you with, uh, you know, using TikTok for example, to as a non, well, I don't know if you'd call it non-traditional, but maybe as a new traditional way to, uh, promote your books?

Keith: I think TikTok is a really interesting. Platform for book publishing. Uh, so book talk is huge, uh, that has really generated the entire new trend around romantic. It is driven it to heights that we haven't seen, uh, for a genre in a long time. And tiktoks algorithm is incredibly good at. Serving that content to people who are gonna enjoy it.

who mention one of our books [: 00:18:00

And what I found was that that. The 20 minutes or 30 minutes at night, uh, before I go to bed, I would always read, and for a while there I wasn't reading. I was, you know, mindlessly scrolling through TikTok and I realized that it may be a helpful way to advertise books, but we're all in competition, whether it's Netflix or video games, or TikTok for people's time, which is.

opportunity and a challenge. [: 00:19:00

Guy: Yeah. Well, uh, interesting point about being, uh, addicted to, uh, just scrolling on, on TikTok and LinkedIn and Facebook. But, uh, yeah, TikTok is just crazy. Um, you know what it can do. So now, now the AI that you're talking about, um, is that mostly, uh, analytic AI or generative ai?

Or, or both. Sounds like most of the stuff you've been talking about has been more analytics.

Keith: Yeah. Analytics I, I think, is the core driver of, of a platform like TikTok in terms of marketing for books, you know, the transformations that we've seen in generative AI have been very impressive and have I.

is just. Drop in the Amazon [: 00:20:00

Now it's down to 10 minutes of revision. It gets all of our favorite words in there and the tone of voice that we like and it, it has just. Rapidly in improved productivity there. Uh, and then with images, with, with product marketing, we can take an iPhone photo of a book and using Photoshop, uh, which, you know, is trained on licensed images, although I'm not entirely sure that's true.

ow, put that into a platform [: 00:21:00

The biggest transformations that we are seeing in book marketing using ai, those tended to be pretty expensive or difficult images to capture, you know, sending an editor out to a golf course to shoot the book, and now we can do it instantaneously using generative ai. Uh, the other big. Transformation that I'm anticipating over the next year is just the improvements to video.

ages. If you look at what AI [: 00:22:00

Guy: Yeah. So how do you see you, you brought up a little bit, uh, going back, you brought up a little bit in terms of who actually owns the, the license or the copyright or of the AI generated image. I've been investigating that a little bit and that it doesn't seem like the legal is there yet. In terms of whether you actually own then the derivative images that AI makes, and, uh, it looks like there's gonna need to be some litigation and legislation and regulation to, uh, to be able to define that.

How do you see that?

uh, in part because it's not [: 00:23:00

Mm-hmm. Even for cover designs, we are not using AI. With some interior photos, we will do cleanup or, you know, my favorite use case for, for AI in publishing is taking an image that's at the wrong aspect ratio. Maybe you need a, a vertical image for this. Book and you have the perfect image, but it's horizontal.

copies and now you're [: 00:24:00

That being said, for marketing, I think that it's a a little bit more open. We are not using tools like Mid Journey or Dolly for marketing content. We are. Pretty much sticking to, to Photoshop, uh, because of the licensing issues. But I know a lot of companies and a lot of publishers are playing around with generative AI for marketing purposes.

Guy: Yeah. Interesting, interesting that you're separating out the, uh, the content side of things from the marketing side of things. Um, uh, because I, I see your point. You know, if you've got a. And whatever it is, a hundred grand in inventory of books and you gotta throw that away because of one sentence or whatever.

That can be pretty expensive, that's for sure.

Keith: Yeah.

hen happening in, um, in the [: 00:25:00

Keith: Uh, well the big thing that's happening right now in publishing is synthetic voice narration, uh, which is gonna be an interesting.

fects of eBooks going back to: 2010

Heady competition. Uh, but what then you saw after that was this price pressure that eBooks put onto paperbacks and then [00:26:00] hardcovers. So we have books that came out in 2010 that were priced at 1495 that were still selling for 1495 today because the market sort of stalled out when this. Price competition came in that ebook just based on inflation, should be a $24 book at this point, and it didn't happen.

Mm-hmm. Uh, audiobook have been this. Pretty wild exception because the list price of an audio book has stayed pretty steady at around $40 because of the narration requirements, the amount of work that has to go into it, and then the way that pricing is structured on, I. Platforms like Amazon, Spotify, and my worry for the publishing industry is that synthetic narration is going to do to audiobooks what eBooks have done to the rest of the, the market and just comm it.

waiting to see what happens [: 00:27:00

And then the other, you know, big. Ongoing development in publishing is, you know, who owns the rights to your content, what can you do with that content using AI or licensing it to companies to train their LLMs on? And how is. That going to change the way that ownership in books is, is perceived going forward?

tly. Um, I imagine that they [: 00:28:00

You know, so I could see, like if I wanted to do something, you know, and go back and, and get stuff from New York Times or whatever it is, the Washington Post, and you know that those are behind the paywall. And so AI may or may not, I don't, I don't even know whether they actually have access to that, that treasure trove of really good, uh, information.

And then similarly, to your point about. Do they have access to the, the full book that, uh, that was written, you know, or, or not? And, uh, I don't know how you guys are, are you putting like gates on the, on those books so that they can't get that? Or, because right now on Amazon, I can get like, you know, the first 10 or 15 pages by, you know, doing the, the read through.

I don't know if, uh, you know, at some point that would be, would be, or could be made available at some, you know, at some cost.

Keith: Yeah, I, I think [: 00:29:00

And that was a huge treasure trove of, hmm. Books, including copyrighted material that no one had licensed out or wasn't in the public domain. Uh, you know, you can't train an LLM on public domain titles alone. You really need a, a huge library to be able to do that. So publishing companies are very wary of.

keep back, uh, the ones that [: 00:30:00

Mm. Who knows how that's gonna go. You know, the New York Times. Got fully scraped by, uh, companies like OpenAI that were, you know, training their LLMs. And that's been a, a big controversy in the, the media market. Uh, you're gonna see the same thing with, you know, videos, you know, Google, which has all of our content, you know, and emails and mm-hmm.

Like Google Docs and YouTube, uh, has a huge amount of data that is available to them. And whether they. Act on it or not is a question that I'm gonna have for a long time going forward. But you know, the multiple LLM that is trained on audio and video and capable of everything is gonna be very hungry for.

Content to, to train on. Yeah.

page, uh, [: 00:31:00

So you have no idea whether that that app is going to take you, you know, has the rights to take all your content and, you know, and feed it to, uh, to an LLM or or something like that?

Keith: Yeah, I mean it's, it's funny 'cause we have authors who are never want their. Manuscript to be uploaded into, you know, chat GPT for analysis or any of these other LLMs that are focused on the publishing industry to do grammar checks like Grammarly or something like that.

nd. How realistic is it that [: 00:32:00

Guy: Yeah. Yeah. Well, fascinating. Um, man, it'll be interesting to see how it all comes out and, and unfortunately, I think it's not gonna come out, you know, in the next, uh, six months. It's gonna be like the next six years or the next 10 years before it really settles down. And, uh, no, it's, yeah,

Keith: it's hard to keep up with, honestly.

Like, it's one of my biggest, uh, stresses right now is. You know, what's changed in AI today and how am I gonna address this with our, our staff and, you know, how is this gonna affect our, our operations going forward? It's just a race.

Guy: Yeah. Sounds like what you need as an AI agent that keeps you up to date on all these AI agents and what's

Keith: going on.

Yeah. Yeah. [: 00:33:00

So foreign rights, you know, an area that is set to. Be completely transformed is, is translation. Mm-hmm. So a translation of a book used to cost $5,006,000. Uh, now you can do an AI translation in 30 seconds and have an editor clean it up for, you know, whatever. They're getting paid hourly. And that has really unlocked a lot of opportunity for moving books from one market to the other at more efficient rates.

Guy: Mm-hmm.

d. One of the things that we [: 00:34:00

And I see a big use case for tools like that in quickly generating marketing content that is B2B. So you're sending it off to a foreign rights agent? Mm-hmm. In Germany? Mm-hmm. The podcast is in German, you know, fully describes the, the book and how great it is and you know, it could be a 32nd clip or a five minute podcast.

know, available for free on [: 00:35:00

Guy: Yeah. It's, uh, fascinating. The, um. I'm actually in the middle of, uh, reading the French book and I don't speak French. And so I've used, uh, uh, the Microsoft, uh, translate, which is only, it's okay. It, um, yeah, needs a lot of work. I mean, I can get the gist of it because I, you know, I, you know, you know the topic and, and you know, you can just get the gist of it 'cause it's not that bad.

But to actually say that it's a, you know, a good translation is, is far from the truth. Absolutely. Yeah. Absolutely. And because sometimes it, it seems like it'll, it'll translate the, uh, the sentence and then sometimes it seems like it's doing it word for word and the word for word. It just doesn't help you at all.

So,

Keith: yeah. No, you, if you were gonna do that, you'd need a good editor to come in and Yeah, yeah. End it up. Yeah.

might as well just work from [: 00:36:00

I, uh. Learned a lot. And I think our audience will learn a lot and where AI and publishing is going. 'cause that's definitely an area where, you know, the, the legal side and the copyright side and the, uh, the rights side is, uh, definitely gonna be a, you know, a big challenge over the next couple years. So, uh, one last question for you.

Uh, what advice would you give, uh, an up and coming new marketer as they're getting into the industry? Uh,

s a highly intelligent, very [: 00:37:00

So the advice that I give to all of our employees across our. Organizations is use it for everything that you're doing. So if you're writing an email, you know, try it out in Google Workspace or whatever. Uh, if you're, you know, analyzing some data that is coming in from your Google Analytics, see what you can do with copilot or with chat GPT, uh, use Anthropic for.

Crafting a, a press release before you go in and, and do it partly because it's gonna increase your productivity, uh, partly because you're gonna find things that you wouldn't have thought of. So we use it for title brainstorming and often it'll come up with a word that nobody had thought of in the meeting, and that is worth.

costs for a, a pro chat GBT [: 00:38:00

Guy: Yeah, that's, uh, definitely, uh, I think the, you know, forcing yourself to use, uh, AI for an hour and just use, you know, as many different platforms as you can and just use them in unusual ways can really help to, uh, improve how, uh, you know, you're, you're just skillset as you're trying to get into the, and break into the industry.

So, uh, yeah. Thank you for that. Really appreciate it. So, um, before we close, how can people, uh, learn more about you and get in touch with you and, uh, where

bout, uh, AI and publishing, [: 00:39:00

Uh, and of course, you know. Connect with me on LinkedIn. I'm always happy to trade stories and talk about what's going on right now in ai because tomorrow it's gonna be completely different.

Guy: Absolutely. Absolutely. And uh, that's kind of the fun of it right now, so, uh, fantastic. Yeah. Thank you for that. I appreciate it.

about my up and coming book, [: 00:40:00

Keith, uh, thank you so much. It was really, really fascinating and eye opening. Thank you for having me. Absolutely.