Home
Sign Up
Nick Torres from Firecrown
Sep 28, 2026
Nick Torres from Firecrown
15
30
00:00
41:56
1×
SUBSCRIBE FOR UPDATES
Subscribe
Listen on
Apple Podcasts
Apple Podcasts
Spotify
Spotify
YouTube
YouTube
Transcript
0:00
We migrated 28 websites in three days. I looked at all the apps that we built, and I tried to put a value towards them, and it came out to be about a million dollars.
0:08
Take AI out of it and just tell me what jobs you hate, and let me help you automate that.
0:13
There are companies who are making more money and are doing things and hiring more people because of AI, and, and I'm happy to say that we're one of them.
0:21
This is Not Just Faster Horses, conversations about deploying AI inside media companies. I'm Ben May, founder of The Code Company, and today I'm with Nick Torres, CTO of Firecrown Media.
0:31
Today we're recording from Firecrown's headquarters in Chattanooga, Tennessee, in probably the fanciest studio, uh, this podcast has been done in.
0:38
Firecrown is a challenger media brand with over 50 websites across 20 different brands in verticals like hobbies, travel, freight, logistics, and more.
0:47
So Nick, you're building AI into every role and every function across the business. Mm-hmm. Help set the scene. What does your product engineering technology department sort of look like at the moment?
0:57
Yeah, so for that many, uh, websites, it's pretty crazy. We really only have a team of four, including myself right now. So from the very beginning, we always, always want to operate a really lean team.
1:07
I think being in a remote world, working on so many websites, having that direct line into the, the dev team is really key, but also operating on so many different websites with so many different passion projects and so many different stakeholders, uh, we've really leaned into being able to streamline everything.
1:25
So obviously you have your ticketing systems, you have your stakeholders for every website. You know, that's all kind of boilerplate.
1:32
But in the last year, uh, we've really migrated everything to a WordPress site, but then also to WordPress VIP, which with Claude now, you have a team of 100 with it backing you. So it's pretty streamlined.
1:44
Um, up until recently, we were on WP Engine, uh, which got us to where we needed to go. But since we're scaling, and if anyone knows anything about Craig, he's always buying new businesses. So it works really well.
1:55
Build a template, drop it into WP, slap Claude on it, keep moving. How have you managed that enthusiasm, uh, from the boss- Yeah... coming down as the s- as the CTO of this business? Yeah.
2:06
Um, and how are you sort of satisfying his thirst to move fast and, and, and do more and more with AI? Well, it's always hard keeping up with it, but I love that about Craig, right?
2:15
I think that's what you want out of your CEO is to not to be complacent, always buying new things, trying new things. We're not afraid to, to buy a new product or try something new and then fail and to fix it later.
2:26
Um, to keep up with him, what we've done, I would say taking kind of a step back, I don't know about you, but it seemed like December of 2025, everyone went on their break and something happened.
2:38
I was on X a lot, and it just kind of clicked for everyone. Leading up to that, Craig and I really started to kind of push AI probably m- mid last year, and it was just slow. Like, all right, what does this mean for us?
2:50
We're obviously a journalistic, uh, forward company, so there was a lot of morality conversations, conversations about, all right, well, is this writing for me? Is it copy editing?
2:59
Like, is this not only gonna replace me, but, like, how much do I need to be transparent about is this written, is it not, right? So we had to kind of fight that, and we kinda got through with it.
3:09
We started to s- you know, roll out Claude. We, Craig and I made the point to say everyone's getting a Claude instance, and my calendar filled up with trying to teach people how to do it.
3:17
But then leading me into that December of '25, and then the beginning of this year, we hit the ground running hard. So like I said, everyone's getting a Claude.
3:25
In the studio, we held multiple town halls where I would have my computer screen up being like, "This is how I'm using it. You tell me how you wanna use it."
3:33
And we found kind of the secret sentence in all of it, which was, "Take AI out of it and just tell me what jobs you hate, and let me help you automate that."
3:42
And I feel like people can relate to the word automate a little bit more than AI. And so from a f- you know, non-technical perspective, that's really how we got people to start using it. And now it's,
3:54
we're heavy users in Replit. People tell me a problem, I throw it into Claude. It connects to a system called n8n, which is like Zapier on steroids, if you've heard of something like that.
4:03
And now we're kind of in a flow where, in a recent acquisition, someone was telling me about a problem in a meeting. I took the Gemini transcription.
4:11
I built a Claude skill that helps me curate something for Replit, dropped it into Replit, pumped out an app within... By the end of the day, it's done and ready to go. Yeah.
4:20
Now, going back to the engineering standpoint, uh, one of my, the, the head of engineering on our end has built countless amount of agents. And what the heck's an agent, right?
4:29
Like- It, it is probably one of the buzzwords of the year and has got a lot of breadth of what it means. Oh my gosh. Yeah. If you don't, if you're not deploying agents, you're not a real CTO. Yeah, yeah, yeah. Whatever.
4:37
But we will say it's just a GitHub repo, right? You have rules. You know how to set them up right.
4:43
But with him and what he's doing, he's having people, like, if we spin up a new WordPress instance, it goes in and drops it and creates it how we want it to. Yeah.
4:51
If we're migrating sites over, we have something that goes in and checks out SEO and URL standards, so we're not cratering there. Um, building a Jira board for every new project, seek an agent on it.
5:01
So we're, we're finding our ways through it. Um, I think we've conquered the what is AI and how to use it, but it's seeming like more people are willing to use it if it's truly helping their job, right? Yeah.
5:13
And I think Craig mentioned in that interview, um, uh, both himself getting in there, building stuff, building apps, things like that, but also, like, I think the story, again, that's stuck in my mind, and there's lots of these stories you hear over the, as you talk to people about, like, I think your HR person, like, built, like, an onboarding app and stuff.
5:28
How, as, like, CTO, do you not end up having, like, your payroll leaked into Slack somehow? Yeah, it, it's funny. The worry of that has not, um, shown and become a fruition, right?
5:41
There's, every system that you're gonna connect to out of the box has permissions, and we've just leaned on that heavy, right? Like, we've set up our own permissions and kinda guidelines.
5:50
I think something that's very utilized, but at least in our world, are skills. Like, we lean heavy into skills, like that Replit skill. Someone comes to me with an idea.
5:58
Before either I build it or they get into Replit and start building it, it's mandatory that you run it through the skill, and our engineering people built that skill. Mm-hmm. So what does that mean? I wanna build an app.
6:10
I toss it in. That is going to, the Claude is gonna work with you to work through these different worries. Like, okay, like, what's the idea?
6:18
Are there any, um, you know, uh, financial documents or, you know, confidential documents? It will ask them questions. Now, of course, you're relying on a human to give you those answers. So then, okay, they've built it.
6:29
How do they get Google connected to it? What are the integrations? That all still runs through us. Like, we still have the admin capabilities,
6:37
but to your point, it's really a balancing act between freedom and creativity, but then gatekeeping that. Not being a blocker as well. Yeah. Correct. That's right. So we're still working through that.
6:46
I mean, HR is a great example, and you're right. She's not only built, um, that specific app for, like, pre-onboarding questions, but all of our career postings are now going through a, a homegrown app.
6:57
Um, she also has, like, an Ask Firecrown. So she puts up her own documents that are public, but like- Mm... simple questions of, "Who do I talk to?" There's an org chart. Or, "What company holidays do we go off of?"
7:09
Everyone can log into- Yeah... their payroll, uh, app, but no one ever does, right? Right. So we're trying... And again, going back to my earlier statement, she hated that job, right? Yeah.
7:18
She hated always having to say, "Yes, we have Labor Day off." Yeah. Or, "Our insurance is Blue Cross Blue Shield," right? Yeah. It's a monotonous task. Yeah.
7:26
So she had creativity to build it, and it's one of our most popular apps now. It's awesome. What are those traits in people that have that sort of part of their brain that- Yeah...
7:35
even thinks to solve a problem that way? Look, and I think it's been awesome. So two points.
7:40
She has openly said, Susie, um, has openly said as an HR person, she feels more empowered and has gotten skills that she never would have learned unless we gave her Replit and Claude and these other things, right? So
7:53
I am appreciative of people, and that's what we look at for when we're hiring at Firecrown. You know, you might be hired at job A, but what are your, um, dreams and aspirations?
8:02
Do you want to dabble in something else, and how can we teach you? That's was... That's my mentality with this.
8:07
Like, I've opened my calendar up, and Susie and I have sit in a room together and be like, "All right," before we had the skills, like, "What are you trying to do? How do you want to build this?"
8:15
And you can see someone get their aha moment, and that's really key for me when I'm trying to instill this. At some point, someone has that aha moment, and to her, we joke she's like a junior dev now, right? Yeah.
8:28
Like, she can get into Replit and troubleshoot stuff. Now, does she know how to set up an API key in Google Cloud? No. Yeah. Does she need to? No. Yeah. But she at least knows what that means.
8:38
You know, she understands a little bit more every day. Yeah. And that has enabled her to start thinking differently. So when she's...
8:46
In a recent acquisition, we just bought the local newspaper here, and she was the fourth, uh, person talking to everyone.
8:53
So when then she was talking to people and on the boots on the ground, her brain has now been trained to be like, "Okay, is this something that we can automate, or is this something that we should build an app for?"
9:02
Um, we actually, when we acquire a company, another app that Susie built is the pre-screen- pre-screening app. So we upload the job requisite.
9:12
Um, a- and she's created a prompt to basically do anywhere from a 3 to 12 questionnaire that's around that job requisite. She could, you know, curate it, adjust it. All in all, it's a typing test. Mm.
9:26
We want to see in the background, like, we're a, a digital forward company. Can you type? Can you do certain things, right? Yeah.
9:32
So it's enabling her to look at her job and be like, "Okay, I'm an HR person, but I can have these tools to do my job better." Yeah. Now, we're not replacing her. Yeah. It's still AI. Yeah.
9:42
So it all goes back to we need humans in the loop to make sure all this is working correctly. And, and in the past, that would've been a three-month project that you would've needed to put engineers on. Absolutely.
9:52
That would've always been deprioritized. Yeah. That's probably one of the most common things in media companies is it's got to be about decision-making- Mm... and we have finite resources.
9:59
Uh, and a bespoke typing test app or an onboarding free just would never- No... it's just not that big of an issue to do, um, which now can be done.
10:07
She's running not only Firecrown, Times Free Press, but, uh, our sister data company, Sonar. She's the HR person for all of them. Yeah. About a 350-person team. Yeah. It's just her. Super power, yeah.
10:18
Insane, but she literally comes out and says, "If it wasn't for Claude and Replit, I wouldn't be able to do this." Not you and the team. It goes back... Absolutely. It goes back to engineering.
10:26
You know, we're running dozens of websites that have dozens of systems that are going through it, and we were looking... And I am hiring two more devs, but a year ago we were looking at hiring a lot more. Yeah.
10:38
But even that says, like, my team feels so empowered to say, "No, I can handle this." How do you think through now the prioritization of stuff, given there could be just so much more?
10:48
It's revenue or reducing cost, right? Yeah. So we have a bunch of things that are falling on our plate. Some may take 10 minutes, and some may take, you know, 10 weeks, 10 months, right?
10:57
But we have to look at that saying, "Is this increasing revenue because it's helping a customer sign up quicker or a better you on a website, or is it reducing costs?" Mm. Those are kind of our barometers right now.
11:09
Um, but what's also cool is my dev team can be running three projects at the same time, and context switching is brutal. It is. It's the same thing for AI. Yeah. You can't toss too much. It, it, it makes that worse.
11:20
[laughs] It's crazy. Yeah. So we do have to have some guardrails there. Um, but also my small team can take on bigger projects, right? Yeah. Because of Claude, we migrated 28 websites in three days. Yeah.
11:32
Because it was just a ve- We set it up once. We tested it. We tested it. We tested it. And then all of a sudden, we're doing 10 at a time. Yeah. And my... Zach was saying, like, "I couldn't be able to have done this."
11:44
No way in the world. Yeah. Or even, like, a full website redesign. Yeah. Something that would've taken two or three people a year to do- Yeah... took one person six months. Yeah. And it was a full-blown- Yeah...
11:52
you helped with it- Yes... um, Sitecore to website data migration. Very old Sitecore to web migration. Insane, right? Yeah. But we have backing.
12:00
And one thing that I don't hear too much about, um, we do put everything into GitHub. So every project is in GitHub, and the reason that we like that, um, is because of the memory behind it.
12:10
So that data migration is an agent now.
12:13
So when we go to another site, it might not be going from Sitecore to web- word- uh, WordPress, but maybe it's just WordPress to WordPress, or we're doing a Django or the Django- Yeah...
12:22
however you want to pronounce it- Yeah... to WordPress. We have the history. Yeah. So like I was talking about the SEO, um, you know, setting up categories and tags and the hierarchy roles, all that's done. Yeah.
12:33
So now we can roll into it, and it might not take six months. Yeah. It might take three. Yeah. It's awesome.
12:36
And that's stuff that everyone's talked about forever in any kind of engineering team is like, how do you codify this knowledge? Correct.
12:41
It's one of the first times that it's literally being codified, and it's all being pulled out by AI. Um, you know, so those rules are getting better and better over time.
12:50
So a lot of those decisions, like the decision frame, that's sort of historically like a product, a dedicated product role.
12:56
Are you now sort of, I guess, thinking about putting your team acro- like, they're wearing multiple hats now- Yeah...
13:01
um, that everyone's got a product lens as well as an engineering lens, or you might have product and design or, you know- Yeah [laughs]... HR and product or like? Well, okay, great, great question.
13:10
It's funny that you say that. So, um, we look at, or I look at my devs almost more like project or product managers, and all of their Claude instances are their teams. Yeah.
13:22
So they're orchestrating multiple different Claudes or agents within it to be able to execute things. And you know, a year ago, um, Zach, you know, was the...
13:34
I wouldn't say resisting it, but just questioning it, like how good can this be, right? Yeah. And took some breaking through, but he's an evangelist now. And is he coding? No, he's not writing out code. Right.
13:44
But he's able to do so much more. A really good example of this is we have our internal entitlement database.
13:50
So when you come to a website, we cookie, we check if you have authenticating or a proper product purchase to read our content.
13:58
That's a homegrown system that he's been working on for a few years, and we got that tech from an acquisition.
14:04
And ever since, even before we purchased them, it's been two years now, they've always wanted to build a WordPress front-end plugin on this product. And it's always, like you said, prioritizing.
14:15
It's not really moving the needle. It should help. Then Claude came in, and he was able to achieve a dream. He did. [laughs] And yeah, he, Claude and him were able to achieve a dream. Yeah.
14:24
And now something that's been on the wish list for years- Yeah... is accomplished. And now- It ended up in production, yeah.
14:29
Yeah, and that's really en- enhanced not only our engineering team, but the marketers, the people, the audience development team are now able to have a homegrown system. They could put up paywalls quickly.
14:40
But the best part about it is, and I, and I love pointing this out, is it was built for us. Yeah. And we control it, right?
14:46
So if, if you're buying something off the shelf, you're dealing with someone who if they have 2X the times contract, their voice is gonna be over you. Right.
14:54
And you might use 20% of what you could be using, but you're paying 100% of it, right? Yeah. So now the turnaround of fixes is so much faster, and it's awesome because it, A, it's a dopamine hit for my guys. Yeah.
15:08
But it's also incredible 'cause it boosts morale. Like an audience- Yeah... development person used to wait months to get something done, and it's like, "No, we can knock it out in a day or two." Yeah.
15:15
And it's really disseminated across the, the organization that everyone be, can be creative, even if they're achieving it or if they're going through us, it's doable now. Yeah.
15:25
There's not a wait time that's, "We're in sprint- Yeah... X, and it'll be three months before we can have a brainstorming conversation with you." Right. Right? Exactly. Like it's, "No, let's get in-" It's done, yeah...
15:34
transcribe the meeting. Let's toss it into Claude, and let's build. Get going. Yeah. How do you... Well, how are you at the moment anyway thinking about what are we building, just vibe coding out?
15:43
Um, what are we borrowing open source or, or buying off the shelf still? Like- Yeah, the build- Have you built HubSpot for yourself yet or, you know? [laughs] We have, actually.
15:51
We have a homegrown CRM, um, that is operating all of our print and digital, um, CRM basically 'cause... All right, let's, quick tangent.
16:01
Forecasting was really hard between print and digital 'cause of the CRM that we're currently using that has to be print enabled, which is a big caveat. It's good. It's a little clunky.
16:11
And so I sat with the sales team, and I built Prototype 1. And going back to prioritization, it worked, but then the CRO came in, and he's like, "Just give me access to Replit. I'm in it every day.
16:22
Just let me fix it," right? Yeah. So now he took it over, and it is his ba- it is his baby. Like we've been- Yeah... demoed other CRMs, and he's like, "Honestly, I like mine better."
16:31
And now is that, is that bias or is that truth? I, I'm tend to saying it's truth. Yeah. Um, but that goes back to the build versus buy, right? Yeah. And it's hard. I say it every day.
16:40
This paradox is just every single decision is really leaning back to can we build this? But I think it goes back to your prioritization key, right? Yeah.
16:47
Like is this gonna be worth the meetings, the, the time that it may take my team away to come fix something? Yeah. Or can we just go buy it? Yeah.
16:57
I think, um, I wouldn't recommend buying or building everything for everyone. I'm, I'm fortunate to have a team who's really embracing it. You haven't thought to, you haven't been pitched to, to vibe code a CMS yet or?
17:09
No, no. Like you're trying to add to the layers where it's- Yeah... it adds value to the business. Correct. You're not trying to figure out a sa- a save button or- Yes... something that already, you know, exists. Yeah.
17:16
Or like, uh- Table stake stuff... I'm definitely not gonna build any like email delivery platform because I don't wanna deal with DMARC. Like there are certain things that could we build something? Absolutely.
17:25
But is it worth the time to do it? Yeah. I think that's where we're coming in.
17:30
But also on the flip side, you know, through this acquisition, we're looking at a lot of vendors, and I'm getting contracts, and it's like, man, like do I really need to spend $100,000 on this, or could I spend two months just heads down and just knock this out?
17:45
Yeah. It's, it's hard. It's really figuring out like what, how much value... 'Cause again, if you're using HubSpot paying six, you know, six figures a year and only using it to like be an address book- Exactly...
17:54
it's like, well, hang on, this is a crazy amount of effort. Yeah. But then I wouldn't also... I've, you know, you hear people say, "Well, I could just build HubSpot in a, give me two weeks, and I'll build HubSpot."
18:00
It's like, well, that's not how it's gonna work. No. It's ne- it's, you know- And there's no one shot. You're not gonna do it with one prompt. Yeah, yeah, exactly. [laughs] It's not perfect out of the box.
18:06
Like there is a lot of time. But then you also have to look, I mean, our, our Claude bill isn't cheap, and our Replit app isn't cheap. Yeah.
18:13
Um, but what we're paying for it is definitely a lot cheaper than what I would have paid for 45 different apps, right? Right. Like even Suzy's- And a, an engineering team of 12 or something to- Correct...
18:24
to do that much output. Yes. Like in all, in all honesty, we're not spending. We're just about to hit six figures between Replit and Claude. For the whole company though. A whole company, right.
18:32
With hundreds of people, yeah. And people are- Yeah... and going back to some people You know, um, building like landing pages, the new Flying Mag, um, subscribe page- Mm... that was 100% built by the marketing team. Mm.
18:44
And all they handed us was an HTML file. Hey, wow. And we were able to kind of piecemeal it. Was it perfect? No. Yeah. Did c- was it a copy and paste into it? Find a couple cut out is how I sort of describe.
18:53
It's like you've got the perfect accurate thing- Yeah... and then your team can just like actually go and build that really quickly. Absolutely. And all that iterating, w- they did it themselves. Yeah.
19:00
Like the, how does this work? How does that work? What if I click this? Yeah. It's all done. And your team just getting like, "Here it is, just pop it on there."
19:05
A lot, a lot of people, and I've pushed this a lot, go prototype it. Yeah. Like, you can build something in Claude that is 80, maybe 75% of the way done. Yeah.
19:15
Then by the time you come to us, again, going back to the communication, it is what you want it to be, so now we just have to kind of push it past the finish line. Yeah. I think another great example is the, the TV team.
19:25
We're, Freight Waves is doing a 12 to 2 daily show, and the first day we started, we hired a social media person. And we wanted to do the, the typical, it's a two-hour show, but you need short form content- Clips...
19:37
the clips and all that kind of stuff, right? So her first day was like, how do we get this person, um, to post this all on s- all on social media, who has no trucking background- Mm... doesn't know the company.
19:48
So we, I sat in a room with her for an hour. I was like, "All right, what's your prior history and how do you do this?"
19:52
And she gave me a few ideas, and in real time we built a Replit app that is now uploading an MP4, gives her a transcription, gives her every social possibility that she can go for. She uploads it to YouTube.
20:06
The kicker is it comes back and now it's writing the article in WordPress. Yeah. Like it's a per- It's draft up ready to go. Yeah... it's perfect. And it's a dra- I was gonna say- Yeah... it's a draft.
20:13
Going back to this is taking jobs. No, human's always in the loop. Like you have to- But humans never liked that job either. Like, are you gonna sit there and type a transcription out or- Yeah...
20:21
copy and paste and like just doing a... Like that, that's the sort of service journalism- Yes... that AI can do. It's like that journalist should be thinking of like the unique stuff- Absolutely...
20:29
that is adding value and not commenting on a YouTube clip. Or watching a two-hour segment- Right... and being like, "Oh, this is cool. That's cool." Yeah. Bookmark that chapter. Bookmark that chapter. Exactly. Yeah.
20:37
Now, going back to build and buy, we started playing around with the idea of should we just build Opus Clip? And it's like, no, no, no, no. That, they're doing great. Yeah. We don't need to do that.
20:46
So- They're gonna be building features for you. Correct. Yeah. Yeah. Correct.
20:49
So it's, it's a fun, it's a fun time to be in this industry and just tech in general, but it is consistently asking that question of build- Yeah... versus buy. Have you had any pressure or to justify the spend?
20:59
M- uh, no. I think I'm in a fortunate position that Craig and I kind of looked at each other at the beginning of the year and said, "Okay, let's give it a year. Yeah. Let's see what happens." Right?
21:06
Um, I definitely don't have a blank check whatsoever. Yeah. But it goes back- You're not tracking maxing. No, no. On, on the board. Uh, yeah. [laughs] No, no, we're not doing that.
21:13
We are pretty stringent on like Claude seats and making sure people... Like, I hold a weekly happy hour where people come in and, you know, I ask them, "I saw that you maxed out your time.
21:24
Like, what e- model are you using? And you do not need to be using Fable 5 for copy editing." Right. Right?
21:28
So like that is a, a part of my job that I love, 'cause going back to that aha moment, you see someone click, it's like- Yeah... "Oh, okay, I understand that." And then they're, they're teaching other people about it.
21:37
But going back to the P&L, you know, the ROI really is, okay, we've built X amount of apps, and I actually did this for a board meeting.
21:46
I looked at all the apps that we built, and I tried to put a value towards them, kind of comparing, right? So like Suzy's HR, the TV app, and it came out to be about a million dollars. Yeah. Right?
21:55
So like that alone shows that the $100,000 that I'm gonna spend this year- Yeah... that's our ROI. Yeah. So yes, there is a new line item.
22:04
Yes, we have to be worrying about tokens if we go to the enterprise account of Claude. Like, there are certain things you have to look at. Yeah. And especially like a good call is like Replit, right? The first,
22:14
the first few months someone's in there, they're building everything, and the tokens do increase. Yeah. So your monthly is a little bit higher. They're building a Calendly replacement, like that's 40 bucks.
22:21
Why are you- Correct. Yes... you know? Yes. But then there's this, there's this moment in time where Suzy spent so much tokens and so much time, or even legal.
22:30
We were paying so much money for a contract management system. Mm. And I said, "Let's just build it." Yeah. And it was a big task, 'cause there's a lot of stuff that goes into legal, to your point- Yeah...
22:39
security and, okay, how is this all integrating? But our legal department, for about a month we sat together, and now it's flawless. Like, it's super easy to be able to run. We would've paid 100 grand for that, right?
22:51
Mm. The reason that I say that is for about six months his spend was high. Yeah. Like, it was high. But now it's just maintaining. It's small little tweaks here and there.
23:00
So you have that like immediate, oh my gosh, you're spending a lot of money on this, but then it kind of tapers off. Yeah. And that's really where you see your, your margins increase, right? Yeah.
23:09
Like your, your cost of building this stunk in the beginning, but long term, it's an incredible ROI because now, again, he's the controller of his fate. Yeah.
23:19
If he wants a new feature, if it's broken, if it's not good enough for him, he's the one that's controlling it. And it's just given him so much freedom to be able to build awesome things.
23:28
Have there been any examples of you being in this journey this year of where you've had to sort of say no or sort of, uh, veto something- Yeah. [laughs]...
23:34
that could have just been like crazy out of left field security costs, whatever that may be? Yeah. I would say, um, the email portion. Yeah. You know, we spend a lot of money on emails. Yeah.
23:44
And we can all say too many emails everywhere. I get it. But there was the idea of, okay, sat in a room for about a week and thought, all right, what's the cost here? And that was a, like, you know what?
23:53
I don't wanna deal with deliverability issue, getting blacklisted. Like, that's just not worth, again, not financially worth it, but just time and effort spent on it. So that was one that was vetoed.
24:03
But we've also tried a lot of things that may have not just come out.
24:05
You know, um, one of the big ones that, that we spoke with, uh, Replit directly on like a customer success story, which was awesome, super appreciative that they called us out, but the first big one was a, an expense management system.
24:17
Mm. So we, uh, hire third-party contributors, and for a while we faced multiple problems, paying them on time, but then also setting a budget and, and staying to that budget, right? Yeah.
24:28
Like we say, "Hey, you have $20,000 for this magazine." Then by the time it goes to the printer, it's too late to realize that we spent 35. Yeah. Where did that happen?
24:37
So- The leakage, and then multiply that across the whole business. Yes. Yeah. And the problem was there wasn't visibility from top down.
24:42
And even at the bottom- You're getting all these throughout the month, and unless that person's super diligent doing a, a boring job keeping a spreadsheet- Yeah... they don't know what's going on, right? Yeah.
24:51
And so we built it, and it's running all of our web- it's running all of our, uh, budgets, and it saved us, like, 100 grand a month- Yeah... after like one look.
25:00
Literally a month, like, it's insane how much- Just the leakage. Yes. Yeah, yeah.
25:03
And so I bring that up because it was, uh, uh, our first time looking into being like, "Okay, what does this mean for us as a newer company?"
25:12
And we could have vetoed it, and then we- And this was probably when you had a lot of new brands coming in 'cause, like, it's one of those ones that's probably a tipping point when there's, like, 30 contractors, someone can sort of keep that in their head of what they've gotta do.
25:21
Yeah. But when you've got hundreds all of a sudden, it's just not humanly possible. Yes. And what's also two factors into that: One, now when we go into an acquisition, we don't need to worry about that. Yeah.
25:30
It's a few button clicks. Um- Yeah... literally with the Times-Free Press, this, the local magazine we purchased, took me 20 minutes to set up their brand, set a budget, train someone, and I'm done, right?
25:40
They're up and running. So then- Yeah... the other portion of that is that connects to the contract system that legal has, so if you enter a new contributor and you don't have a contract, okay, we can't pay them. Yeah.
25:51
So you upload it, connects over to the CRM or the PLM. Ticks all the boxes. Uh, yeah.
25:55
Yeah, like, it, it enables not only internally morale to boost, but now our contributors are als- are stoked because they're getting paid on time, right? Yeah. Like, it, it...
26:03
There was a trickle-down effect going back to your original question of ROI. Craig and I, and really the executive team in general, realized that it might not be dollar value.
26:13
In this case it was, but in other ways, just helping our business run smoother. Yeah. And we need to be lean. We're a media company. Yeah. Everyone knows, you know, profitability's tough to come around.
26:23
Existential crisis- Yeah... since the dawn of the printing press, you know? It's, uh... Yes, page views, we're fighting for every single one of them and- Yeah... and we have to stay nimble and lean.
26:30
And so- And abil- a- ability to pivot fast. Correct. Like, that's the competitive advantage if you have a six month to turning it around versus six days- Yeah...
26:38
then you can actually respond and- And a lot of the things I feel also that aren't being talked about enough are people who are embracing it, and the fact that we've really given the keys to the castle to everyone.
26:49
Of, of course, this isn't 100% across the board, but I do see people enjoy their jobs more, right? Yeah. Like, I do see- Well, you're, you're giving them a pathway to get rid of the crap parts of their jobs. Yes.
26:59
Like, "I hate doing this every week." Yeah. And now they can actually... There's a pathway- Yeah... a viable pathway. And they're feeling like they have a little bit of stake, and they do have.
27:07
Like, um, another version of the local newspaper, um, the sports guy, the sports editor, was telling me that they would get, uh, stats on every high school game via email, and then they would print it out or copy and paste it.
27:19
Like, just monotonous tasks, right? And so we built an app, and it's a lightweight app. Yeah. I'm not reinventing the wheel here. Yeah, yeah. It's a login. It's a roster. Yeah. It's a data sheet.
27:28
Goes to something else, right? Like, these allow people, and now what would've taken him four hours out of his Wednesday should take him maybe an hour right now. Yeah. It's not instant. It's not perfect.
27:37
There's gonna be hiccups, but even he, that was his aha moment. Yeah. And I think I try when I talk to others- And it just kind of compounds. It does. Like, it just gets more and more and more. Now I have- Yeah...
27:46
and going back to your point, prioritizing. Yeah. Now he has two more ideas. Yeah. And so now it's like, "Cool, let me train you on Replit so I'm not that gatekeeper." Yeah.
27:53
And again, I think to answer your question in more detail, like, we're not... Most of the apps that are being k- like, are brought to us, they're not P&L describers. They're not payroll man...
28:06
Like, we have a commission flow that's locked down. But even at the end of the day, I'm not uploading salaries. Like- Yeah...
28:11
the worry that a lot of people have with AI, um, is valid, but I would say don't put so much weight on it until it presents itself, right? Like, of course you need to prepare and you need to have guardrails, but
28:24
it's not like people are trying to, like, tap into the mainframe of our database. Like, we're a media company. Yeah, yeah. [laughs] I'm not a government entity. Like, I don't have too much to worry about there. Yeah.
28:32
So it has been... That's another thing that I'm fortunate to say, is that I'm not, you know, making bombs over here. Right. Exactly.
28:39
It's just a couple of copy- There's no secret, secret data that, uh, yeah, the NSA locked down or something like that. Yes. And that happens with engineering too, right? Yeah.
28:45
We're not making anything proprietary, you know? I mean, that's always the most obvious thing when you talk to media companies about tech. I'm like, "The actual core business is incredibly simple." Yes.
28:54
"You're putting words on the internet." Yeah. Like, it's sort of a solved problem. [laughs] Yeah. 20 years ago, like the dawn of the internet, it was a solved problem. Yeah. Um, it's not overcomplicated.
29:01
And with that, though, something that was another app that we built is we have a lot of print editors that have a hard time really grasping what a SEO, GEO, whatever three-letter acronym you wanna throw at them. Yes.
29:14
And so after trying so many times of installing Yoast or installing all these things that should have helped them- Yeah... the learning curve just isn't gonna work, right? Yeah. So we just built an app- Yeah...
29:25
that they, again, going back to the morality of a journalist, you're writing the content, but before you- They never like doing the SEO stuff- Yeah, they hate it... or the alt text or the, the- Yes... the metadata.
29:34
Exactly. Like, all these things that are- So they write it and whatever they write. Before you publish- Yeah... just drop it into this app. Yep.
29:39
And every month it's looking at, um, it's refreshing its memory to understand- Mm... the newest and latest, 'cause that's also- Important, yeah... it is someone's job to look at that. Yep.
29:47
But we have a helping hand now, and that's been incredibly helpful because now, again, going back, the journalist doesn't have to think, "What's my headline?" Yeah. Give them five. Yeah. Pick which one.
29:57
Give them the slug. Give them keywords. Yeah. They're back to writing. Exactly. Like, it's, it's not as hard to look at a small win as a big win- Yeah...
30:05
especially when things are still kind of broken in the media industry right now. Yeah. Um, so it's been fun trying to kind of piece things together. Yeah.
30:12
But at the end of the day, it's not like I have to put up 8,000 firewalls, right? Right. [laughs] So you're... Be creative. Yeah. Have fun. We'll figure it out after it happens. Yeah.
30:20
Uh, how are you thinking in your sort of role in advising and all of that with the business how AI, uh, is gonna help, I guess, transform a, a really traditional legacy business? Where do you start even? Yeah.
30:31
All right, so, um, where do we start? Same thing as everywhere. I asked every single reporter. We have happy hours, and we have... town halls, but what do you hate about your job?
30:40
Because that's my first question of, like, 'cause they're traditional journalists. Mm. We- we've beaten the dead horde on- horse on this, but they have morales. Yeah. They don't wanna use AI in certain areas.
30:50
So I said, "Okay, I'm not gonna win that battle today." Yeah, it's not gonna be AI content this week. Uh, yeah. No, it's not writing for you, but give me something that I could show you.
30:57
And so one of the courtroom and crime reporters told me that every morning he wakes up and he has to look through so many different sources to find arrests. Mm-hmm.
31:06
And most of them, they're not gonna move the needle on page views.
31:09
But he has an accompanying list that has names or has, uh, a scoring metrics that we came up with that if it's a certain kind of crime- Notable [laughs]... notable figures... notable arrests.
31:20
[laughs] Yes, it should shoot up, right? Yeah. And so I said, "Cool, perfect. Let's work on that." So sat with him for an hour.
31:26
We went back and forth with Claude, and about a week went by, and we were still kind of tinkering with it. Then we had our first AI, like, town hall, and it was like teaching a class. No one wanted to talk.
31:37
I understand they were processing it, too. Yeah. But an aha moment that happened is I said, "All right, Mason, let me show you what I built you," in front of everyone. And I was scrolling through it.
31:46
He was giving some feedback, and then I just decided it's in N8N, so it looks like a chart. Nothing pretty.
31:52
I scrolled all the way to the right, and the last column was updates on this case, and he interrupted me and went, "Whoa." Yeah. "I've never seen that before." And then that's his aha moment, right? Yeah.
32:03
Like, he went from "I don't think you're gonna be able to help me" to, "You know what?" He literally said, "I woke up today thinking there was nothing to write about." Yeah.
32:10
"And now I know there's something at 1:30," right? So, like,
32:13
that kind of gives me chills, but it's like it's so cool to see someone get that aha moment because now, again, like the sports editor, he has so many different ideas to come to, and it's helping him- Yeah... write.
32:25
That's what he wants to do. Yeah. He doesn't want to be a clerk looking at all these different cases. He wants to write, and we're getting him back to do that. Yeah. So that's one.
32:32
The data aspects, you know, there's a, a s- How many years of archives have you got? Oh, man. Like, I'm guessing there's a ton at IP. So much.
32:38
And even going back to the, the crime reporter, like the ability to within a second understand, like, patterns over years or- Yes... you know, the, that that store's been robbed 30 times over the 50 years- Yeah...
32:47
or, like- Yeah... to do data analysis that a human just doesn't have the resources or time to. Yeah, so and that, that is something that we're looking to roll across our entire infrastructure.
32:55
So they do have hardware, which is crazy, and so that's something else, um- A lot of opportunity or, or- A lot of... Yes, absolutely. We haven't hit... We haven't s- you know, put Claude on it yet. Yeah.
33:05
[laughs] That's our next... Like, we were joking. They gotta get some tokens and- Yeah, like how do you... Like, if looking at your s- your server, it's almost like, okay, Claude's here, not here yet.
33:14
Let's just get him in, and we'll see what happens. But it's been an awesome experiment. It's been an awesome company to purchase.
33:20
And another thing, they have so many copy editors there that they're, they are overwhelmed with the amount of content that they have to copyright. Mm. And so they were worried, like, how do I make sure that it's right?
33:31
So I built a skill that is the copy editor prompt. So now they're able to go in and say, "This is how I copy edit." And so when they drop comp, uh, a, a situation in or a story in, it's like having a...
33:43
I always say it's like a new hire. First couple of times, you're training it to be how you want the new hire to be. But hopefully two or three months down the line- Mm...
33:51
it's, they're ready, and you don't have to worry about handing them a piece of work, and they know how you work. That's exactly how we operate, and they're loving it.
33:58
It's awesome to see their eyes grow and be like, "Cool. I can do five times the amount of work, but I'm not stressed." Yeah. It's all working. Like, it's all happening in front of my eyes, and it's an awesome thing.
34:09
The in- the UI is super cool. Yeah, right. But it's a, it's a cool premise to be able to show them that we're trying to help you out as much as possible.
34:16
Something I was just thinking about, and I don't think I saw any coverage on AMO about this, but when you look at buying something like a newspaper, would you have done that a year ago without AI for the mix?
34:26
Because you think of, like, the technical debt and the amount of work. Like, if, if this was happening a year or two ago, you would've had to send 20 people in- Yeah...
34:33
to do this amount of work and this uplift to bring it up. Yeah. Like, it's an interesting thing I was just thinking through then of. Yeah. So, um, uh, a two-part answer. Yes, we would have because that's Craig.
34:43
Well, true. [laughs] He buys things. We buy things. You'd have figured it out. Yeah, that's right. Yes.
34:46
The second thing is we have a 90-day TSA period, which means that the, the company that we bought it from had been awesome, but w- we need to get out of their home, and it- Yeah... sits in 90 days, right? So
34:57
the biggest, craziest thing is we are not getting their design front end, and we're not getting their back end for the CMS. Mm-hmm. So we had to...
35:06
In a year ago, it would've been me and Figma doing it by hand, which- Yeah... it stings. Yeah. [laughs] If you know Figma, great, but if you don't, it stings.
35:14
And then standing up a WordPress site and getting all that done, not the hardest thing in the world. Yeah. Pretty templatized.
35:19
But leading up to announcing the acquisition, I spent two weeks redesigning the website using Claude and a Figma MVP. Yeah. So by the time we announced it, I had everything. You already did, yeah. I had all their pages.
35:30
I had everything, and I improved it. Yeah. Right now it looks a little bit better. It's a little bit more cutting edge. And so to answer your question, it would've been drop everything you're currently doing- Yeah...
35:40
and move over here, my three devs. Probably also... Yeah, exactly. Probably need more resources given- Correct. Yeah, exactly. And now it's the design's done. Yeah. Jeremy, another dev, is standing up the WordPress.
35:49
Next week we'll test it. Yeah. And we're 45 days ahead of schedule, right? Yeah, yeah, exactly. It's an amazing tool to have.
35:54
But, but 90 days, 'cause again, I've seen these sort of M&A deals before where you'll do, like, a 12-month, like, lease back on- Yeah...
36:00
the infrastructure 'cause, like, there's no possible way we're gonna move a jewelry platforming in 12 months- Yeah... for a legacy brand or something like that. But- Yeah...
36:07
yeah, if you're, or if you're at, like, beta in 45 days- It's quick. Yeah. And, and now the ed- editorial team can come in and have ample time to test. Mm.
36:15
Where before we would've just been like, "Hey, you're getting it and run." We're just ripping the Band-Aid and just, just- Yeah. And there's pros and cons of that, but it allows us to go through.
36:23
The other thing that I think, um, I haven't really heard about, when we were going through the M&A, we got very little information about it.
36:30
And the information that we did get, plus the meeting, um, transcriptions, we created a Claude project for.
36:36
And again, not personal information, but it just allowed our remote executive teams to decentralize it all coming off- Have, have an know what's going on, yeah... absolutely, going off the same data.
36:45
But now what's crazy is now that we're in it and we're three weeks in, that project has so much context behind it, especially with, like, the hardware moving over. Yeah.
36:54
So as we go through these vendor contracts and all these decisions that we have to make, we kind of have a guide rail of Claude being like, "Well- "You originally said this. Do you really wanna go that route?"
37:06
Or it knows that we currently don't use Auth0, but this company does, and actually worked through a whole plan with us, being like, "It actually makes sense if you stick with their Auth0 'cause then you can migrate more of your company."
37:15
Like- Yeah... it is truly an amazing tool to have, and w- we do do a lot of acquisitions, and this was the first one that Claude had its, its hands on it. Yeah.
37:25
But it was great 'cause now anyone can come to that project at an executive level. Yeah. And it's also scanning, uh, emails, so it's updating in real time. Yeah.
37:34
We hired someone to be kind of like the, the, the consultant on this, just making sure everything's going through- The lead of it. Yeah... yeah, right?
37:40
And we have a spreadsheet that every night it scans through multiple emails, and it flags things, right? 'Cause communication- Yeah... we're getting a lot of emails. Happening in a lot of different places at once.
37:49
Yeah And we all can't be in the same meetings also, and I think that is such an incredible aspect of our company is if you have Gemini recording on it- Yeah...
37:59
and you have Claude look at your recordings, you can start to piece together things that may have not been communicated, especially in a remote world. Yeah.
38:07
And so I don't have to be in every meeting, and my consultant doesn't have to be in every meeting. But every morning we wake up, we get a TFP email that's like, "Hey, guys. What's the plan here?" Yeah.
38:16
And now it's like we are able to hit deadlines- Yeah... with not working 80 hours a week, right? Right. Like, I still have a weekend, which is incredible. Yeah. Where the last acquisition, I had no weekend. Yeah, yeah.
38:26
[laughs] I also have a newborn, so I got an eight-month-old. Keeps, keeps you busy. Yeah So I have to have one, but yes, like- Yeah...
38:30
it's an incredible tool, and I, and I will, I will talk, I will talk to any company or any CTO or anyone about, like, how can you have this amount of memory and context- Yeah...
38:41
and have AI help you out 'cause it's not perfect always. Sometimes it spells something wrong, and you're like, "What the heck is this?" Yeah. Or the, you have to have a human in the loop. Yeah.
38:49
But it is really helping us- Yeah... become a faster horse absolutely. Yeah, exactly. What I guess is exciting you the most about where this is going?
38:55
Like, what is the, the, the longer term where your head's at when you get out of bed? Yeah.
38:59
I think, um, more acquisitions are coming obviously, and, and I think the cool part about it is at my position, I'm able to, again, I go back to that aha moment. You know,
39:11
I do appreciate that AI is being adopted, but I still, still see reservations in a lot of people.
39:16
So from more of an existential point of view, I'm excited to reach more people and get them, uh, excited about AI and kind of break down the, "Oh my gosh, this is gonna take over all of our jobs."
39:26
And it is gonna reduce some, but you know, my dev team embraced this, and they've actually said like, "I feel more confident being a developer right now because I can do more. I can- Yeah...
39:37
I can have more control, and I don't feel so overwhelmed of the projects that are getting thrown at me." Yeah. That at, at a more morale and more, like, human-to-human aspect- Yeah...
39:48
I love, and I'm mostly excited about that. From a more internal capability, um, one thing that we're really working on is centralizing data. We have so, we have- Everyone has so much data that they don't re... Yeah...
40:00
tens of hundreds of millions of records to try to figure out how we can better be, um, positioned to provide the best content to certain readers and, and not just be a run-of-the-mill media company. Yeah.
40:11
I think that's really where we're going towards.
40:13
But I'm starting to get into, like, data engineering and understanding all of that and, you know, understanding that, you know, bad data in is bad data out, and that's something that we're starting to understand.
40:24
There's a lot of companies coming my way like, "Oh, we can do that in a couple of weeks," and it's like, I, I would love to see it- Yeah... if you can. Amazing. I'll give you all the money in the world. Yeah.
40:32
But also, like, it's just a lot of data. So many assumptions. Yeah. It's so much, and again, AI's not perfect. Yeah.
40:38
I, I will be the first to tell you there's a lot of times where it's like you should probably check that data or check that, you know, quote. Yeah. They- Yeah... stinks with quotes. AI isn't replacing all the jobs. Yeah.
40:48
It should be looked at that it's helping you do your job better, which is giving your end user, whether it's a reader, whether it's a client, whatever the case may be, it should give them a better outcome too, right?
40:59
And I think, I am hoping that conversations like these could really point that out because, yes, you're gonna get AI slop. Yeah.
41:07
You're gonna have the Will Smith eating spaghetti video that everyone is so crazy that's the barometer of an AI video. Yeah. But it is.
41:14
But you're gonna eventually get to the point where it's like, okay, the team is still here. We're still doing work. It may be more work- Yeah... but it's work that everyone's enjoying, and it's- Yeah...
41:25
allowing revenue to increase, so maybe I do go hire more people, so I can get more jobs. Like- Yeah... there's the, uh, ev- if it doesn't, you know, the doom and gloom's gonna- Yeah...
41:35
always be the highest, you know, value of, all right, let's go check out and read that, and talk about it, and get clicks and views. But I wish and I hope that there's more happy stories like that out there. Yeah.
41:44
There are companies who are making more money, and are doing things, and hiring more people because of AI, and, and I'm happy to say that we're one of them.