Home
Sign Up
Scott Purcell From Man of Many
Nov 2, 2026
Scott Purcell From Man of Many
15
30
00:00
28:57
1×
SUBSCRIBE FOR UPDATES
Subscribe
Listen on
Apple Podcasts
Apple Podcasts
Spotify
Spotify
YouTube
YouTube
Transcript
0:00
It's crazy to think that we built this AI system back only in February of this year. Mm-hmm. But it feels like I've lived 10 lifetimes [laughs] in that time. Two years ago, on a Tuesday, probably death by emails.
0:15
It is quite dangerous, like we don't want someone to kind of all of a sudden delete everything on WordPress or things of that nature.
0:21
There's expectation that you just produce more with less people, and I don't think we're going down that path.
0:27
This is Not Just Faster Horses, conversations about deploying AI inside media companies across people, product, and technology. I'm Ben May, founder of The Code Company, and today I'm with Scott Purcell from Man of Many.
0:39
Scott, you were a financial analyst at Westpac when you first started Man of Many in 2012 with your co-founder, Frank. Yeah. 14 years on, you're still a two-founder business.
0:48
But when we spoke earlier this month, no, earlier last month at Mumbrella Publish, you said, uh, after the talk about Not Just Faster Horses, uh, that, you know, you think you've built the car.
0:58
What does Tuesday look like for you now that it didn't two years ago? I think my entire relationship with work has completely changed with AI.
1:08
Like, it's crazy to think that we built this AI system back only in February of this year. Mm-hmm.
1:15
But it feels like I've lived 10 lifetimes [laughs] in that time, just with the amount of change and rate of change of, of things and how they, how they're kind of going about. I think, um- And you're talking about Otto.
1:27
This is your- Yeah, yeah... chief of staff, uh, you've named. This is my A- our AI operating system that kind of holds the glue together the, in the company in terms of- Yeah...
1:37
um, all the metrics and reporting and, and insights that we get out of the business. Um, which I'll, I'll explain a bit more later. But I think two years ago, on a Tuesday, probably death by email.
1:50
Spent [laughs] a lot of my time bogged down- Just fighting fire. Yeah. Yeah. Bogged down in emails, trying to find the right files, a lot of manual work.
1:58
Um, I think what's changed now is that, um, I'm able to do a lot more of the important work and have the information delivered to me in a way that I need. Mm-hmm.
2:08
Um, and I've helped AI kind of build and craft a lot of that for me so that I'm not having to go looking for information. Like, I can chat with the data now.
2:18
Um, all of the reports are going to each of the team members, and they're able to action them rather than having to spend hours kind of producing those things.
2:26
So Otto started, I'm guessing, as an experiment, as a lot of AI, uh, um, ideas are. When did you start to think it was becoming sort of a real thing? Yeah. I think, um,
2:36
we s- we had a million different experiments with AI. Like, I looked back in history, I think you were there at a presentation I gave where we had an on-site AI chatbot way back in 2023.
2:49
Um, and then we moved into things like custom GPTs and gems. Um, we used Google, like, Workspace kind of automation studio, um, to look at automating a few workflows and things like that.
3:04
Um, then I started following a few YouTubers, Nate Hurk and Nick Sarayev, who were really helpful to me. They were leading [laughs] into a lot of, um, automation with n8n, and building those was quite complex. Mm-hmm.
3:19
Um, and then quickly as kind of agentic AI workflows came about and I realized you could build those complex workflows simply by chatting with an AI and let it build it for you, that was kind of like the, the light bulb moment going off that I can actually explain all of these automations that I wanna build, and AI understands,
3:40
um, in plain English what I'm trying to achieve, but can set up kind of that technical backend of some of those linear workflows.
3:47
So that's helped us enable automation from, um, a lot of briefing for our editorial team, um, a lot of reporting, um, and then also a lot of tracking and alerts and stuff.
4:00
When did it sort of flip from being, like, a series of automations to, you know, I guess the whole framing of Otto now as a, as a product, right? That's where it's landed. Yeah.
4:09
I think we started with pretty simple automations, like, um, being alerted when a new post was coming on Instagram or reminding the team to like stuff. Um,
4:21
where we started to get better into it was h- giving AI more context in our business. Um, I'd build like 40 different notebook LMs, and trying to get insights out of them was quite manual and tedious.
4:35
But then I quickly realized, like, if I put all of that context on a server and gave it to Claude- And centralize...
4:42
um, yeah, and centralize it, um, it was able to deliver me much better insights and answers back to me than, than even I could. Mm-hmm.
4:50
Um, 'cause it had all of that history, all of that context in the business and it's just, uh, I use it now basically as a giant search engine for our business.
5:00
Um, and it's able to pull in the right amount of context that it needs to kind of answer some of those questions.
5:05
So how do you explain to someone maybe less technical or, or, or even somewhat technical, like, what is Otto? Um, the best way to kind of describe it is it's an AI operating system for our business, in simple terms.
5:18
So what that means is we've got a virtual private server, which is essentially a comp- computer in the cloud. Mm-hmm.
5:25
Um, and to simplify it, we have a bunch of folders, which are things like our knowledge, company knowledge, um- Our, uh, automations are all loaded on there, our scripts, um, things like, uh, our company wiki.
5:43
Um, and then basically using all of those different folders as context, we're able to build automations off the back of that.
5:51
Um, and that computer that sits in the cloud is synced up to all of our different systems and processes using APIs or MCPs.
6:00
So are you storing things like knowledge and processes and things like that, sort of in your own system, and then connecting to a, an LLM like Claude after that or? Yeah, it's, it's pretty much model agnostic.
6:11
So, um, Claude Code sits on top of all of those folders, and we use that to kind of interact and chat with all of the data that we've stored, stored. Um, but we could swap in any other model.
6:22
So I've, I've tested that, um- Mm-hmm... and been playing around, um, a lot lately with other different models.
6:29
Um, sometimes [chuckles] I'll exhaust my context and have to switch to another one, but it's really helpful because we own and retain all of that data. Like we're not- Yeah...
6:38
reliant on having to be, um, within any one platform because we've built it that way. So, um, you can kind of swap and chat with any model on top- Yeah...
6:48
so long as it's kind of secure and set up, um, in the right way as well. Right. And how are you thinking about, like, what a, a lesson is, um, as Otto gets smarter?
6:56
It's a bit of a combination of both, and I think a lot of people have built some really useful, um, GitHub repos around self-learning. Mm.
7:04
So, um, we've got a number of kind of self-learning loops that look back over the previous week.
7:11
At times I've corrected the model, so might give me an output or an answer, and I'll say, "That doesn't seem right," or, "Actually, I want it in this format." Mm-hmm.
7:19
And where I've told it multiple times to perhaps correct the same thing, it'll, it'll report to me at the end of the week and say, "Hey, I've noticed you constantly correcting me on this.
7:29
Do you want me to update this moving forward?" Um, so I'm still in control of it. Mm-hmm. It's not necessarily auto-updating. Yeah. Self-healing. Um, yeah, self-healing or so.
7:39
It also tells me, like, if something breaks, like often sometimes a cron might not run, I've exhausted some API credit somewhere. Yeah.
7:46
But it's still helpful to kind of know that it hasn't worked, and then I'll kind of chat back and forth with it about how to fix that, how to get it working again.
7:54
Um, I think there's self-learning from a business perspective as well. Like, I, I used to be bogged down and spend hours and hours reading newsletters and updates- Yeah... and constant, um, constantly...
8:07
'Cause I love learning about all of these things and, and the publishing industry in general. So,
8:12
um, being able to put that in a format where it's really digestible for me, but have all of that context still sitting in, in that database online- Mm...
8:20
is really helpful, 'cause I can kind of draw upon that and ask about what the latest industry updates or news are when I need them. Yeah.
8:27
As you think about all the sort of small individual problems that you've fixed throughout the business, uh, I'm guessing it's moved beyond just your sort of workload.
8:36
How is it sort of impacting the broader team at the moment across, like, editorial and commercial and things like that? Um, we've been pretty open with the team about, uh,
8:46
giving us a wish list of things that, that, that would save them a huge amount of time. Mm-hmm. We haven't opened up the kind of virtual private server fully to the team, so they don't have their own- Yeah...
8:57
instances of Claude and things, 'cause there is some... Because it is connected to everything, it is, like, quite dangerous. Like, we don't want it to be...
9:06
We don't want someone to kind of all of a sudden delete everything on WordPress or things of that nature. Yeah. So the only- Yeah, there's a lot- Yeah [laughs]... a lot of powerful stuff in there. Yeah.
9:14
Yeah, so the only people who have a- access to it currently are Frank and myself, but what we've done to still allow the team to use these things is to build, um, an online dashboard. Mm-hmm.
9:26
So they can have access to the tools that we build for a specific use. Yeah.
9:31
So an example, um, might be, like, a social caption generator, um, where we've taken a bunch of learnings, like I mentioned earlier, around what sorts of captions or hooks or things perform on social, looking at our analytics, um, building that into kind of like a, a model, and the team can go on there and say, "Hey, I wanna do a post on the new James Bond Omega watches.
9:54
What's a good catchy hook?" Yeah. Um, and then back to that human in the loop. AI's not publishing that verbatim. Yeah. They have to use their own judgment about, like, "Is this a good idea? Do I wanna reword it?"
10:06
But it still tells them and helps them inform them with what the data has said over the past four years around what's gonna perform for our audience and stuff like that. Yeah.
10:14
How have you thought about trying to flip that from steps in a process to, like, a system, I guess, of parts of the business where, you know, you're sort of combining lots of parts of the roadmap?
10:24
It's, we've moved a little bit away from end-to-end, so a lot of the, like, automations now sit within Claude Code. So Claude Code will build a script or a cron- Mm-hmm... that will run on, on a schedule or a trigger.
10:37
Mm-hmm. Um, I'm thinking of podcasting is a really great example where there's a bunch of steps involved in that with, like, filming, editing, promotion. I can relate to that. Yeah. [laughs] There's a lot of steps, yeah.
10:50
So when we think about that as a system, we think about what parts of that can we automate. Mm. Show notes is a really, really good one- Yeah... 'cause we've got, like,
11:00
we wanna have, like, a consistent format each episode that we publish. So we've used AI to kind of automate that formatting. The team can just simply copy in the transcript, and it'll spit it out really easily for them.
11:12
So that's, like, one part of a system that we've automated. Mm. But there's still obviously humans doing most of the work in the podcasting. Yeah. It's just looking at those
11:23
time-consuming tasks that aren't necessarily, like, when I say, um, high value, I'm, I mean, like- High effort of brain, but ti- very time-consuming Yeah So, writing show notes is not difficult- Yeah...
11:37
but it's just very time-consuming A lot of time-consuming. Yeah, yeah Yeah.
11:39
So those are the things that we're really looking to try and automate, and then we wanna think about, um, ways where we as a team can add the most value to that process, I think.
11:49
How do you get a broader team involved in what to automate next and put through AI? We try and do it pretty quarterly, so I know you're also a fan of the EOS- Yes... [laughs] um, Entrepreneurial Operating System.
12:02
So we have, like, weekly meetings with our senior leadership team. Um, usually as part of that, we'll discuss, like, what automations we're currently building or- Mm-hmm... any requests the team might have from me.
12:14
Um, and then I'll kind of work with them to make sure it's- we're building those tools in a way that works for them. Um, usually it doesn't get it right. Like, we'll have a few iterations and tests internally. Mm-hmm.
12:27
Um, as a general rule, we've just recently updated our AI usage policy, and there's a lot of really important stuff in there.
12:34
Like, people talk about human in the loop, and we're really careful not to have AI publishing- Yeah... any content direct to our site. Everything has to pass through human hands. Like, an AI can't send an email.
12:46
It still has to be sent by one of us, um, and edited by one of us. Yeah. And they're similar to the kind of...
12:53
Google, just over the weekend, released some new guidelines, um, saying that all AI-generated content must be fact-checked by a human- Right... um- Yeah... under the SEO helpful content guidelines.
13:05
It'll only get tighter and tighter. Like, there's- Yeah... this weird gap at the moment where it's not explicitly banned, but- Yeah, it was just- Will only go one direction.
13:12
When you're thinking about what tools and systems to automate, how are you sort of choosing what gets, makes the cut, what, versus what doesn't? It kind of comes back to that kind of, like, product management.
13:22
Also, like, each of our senior leaders look after different divisions.
13:27
So our, we've got a pretty small team of about 15 people, so we've got sales, branded content, who look after, like, our sponsored content campaigns- Mm-hmm... um, editorial, and then social.
13:39
So across each of those, um, the senior leaders have to justify how this automation is gonna either save money or save time. Mm-hmm. Um, and that's how we'll kind of look to prioritize those things. Um,
13:53
also, like, the, and we'll use those senior made, senior leaders meetings to kind of help prioritize those things as well. Are you trying to sort of embrace more, I guess, product thinking?
14:03
Yeah, I think the iteration's been so much quicker. Like, two really good examples was, um, we've just automated some of our, like, Facebook posting and LinkedIn posting. Mm-hmm.
14:15
Um, just testing out some new formats, um, with images and some, a link in the comment- Yeah... and, and standardizing some of those images.
14:22
So previously we'd have to go into Canva, create the image, chuck in the- Yeah... text. Yeah. Now that's all, it's all handled by our operating system essentially.
14:33
Um, and we knew within a week what the impact of, of that was just because it's also got access to the analytics. Mm. So it was just a matter of me saying, "Hey, we've been running this.
14:43
Can you let me know how it's going? Can you graph it up?" And it was really easy to see the uplift and outcome from that project, um, within a week, so we've, we've kept it.
14:55
Um, but it's really important that you can use these analytical tools off the- Yeah... off the back end of building them- Yeah... um, that they're actually making a difference. 'Cause I think
15:05
in the initial stages, I was, like, a bit, um, what's the word, tool-heavy, where I was building a tool for everything and then not using it. Yeah. I'd have, like, dashboards and- Yeah...
15:19
metrics and [laughs] Well, you get that first thing so quickly, and then, like, the fine-tuning actually takes 80% of the time. Yeah. Yeah. Or I'd build...
15:25
I, I got a bit dashboard-obsessed where I was building, like, analytics dashboards for every single- Yeah... different part of the business, and no one used it. No. [laughs] Everyone's always- And-... wanted dashboards.
15:34
Yeah... so I think now it's, it's, um, we recently just refined, like, each of the KPIs that the team looks after. I think it's, like, two or three numbers that they each own and- Yeah, okay... are responsible for. Yeah.
15:45
So it's, they don't have to worry about, like, they still have access to those dashboards, but- Yeah... they're not really actively looking at them. They're worried about those, those kind of key numbers. Yeah, right.
15:54
Um, how are you sort of thinking about ROI? Like, obviously you're really passionate about it. Who's sort of the person in the room who says, "No, uh, this is more of a pri- something else might be more of a priority"?
16:04
[laughs] Is, is Frank that sort of, uh, yin and yang or? Yeah. I get enough pushback from some of the team. Like, some of the team hate AI- Not, not as AI-pilled as-...
16:13
in the sense, yeah, not as AI-pilled as probably I am, I would say. Yeah. Like, and I think I, I've always been a bit of a skeptic myself with most- Yeah... most stuff. I think, um, it's good to have those detractors.
16:23
Like, I never wanna have yes people in the team. Yeah. Like, it's good to have people pushing back on, "Hey, I think this- Yeah...
16:30
has gone a bit too far," or this- Well, at least makes you think about it yourself a bit deeper- Yeah... 'cause you can sometimes just, you know, au- auto yes to everything and then- Yeah...
16:37
realize, oh, wow, yeah, I'm actually way down a rabbit hole. Um, like, speaking honestly, like, I, a, a thing that I am guilty of is, um, having sent really long AI emails in the past- Yeah...
16:50
like, and the team's just like, "I don't have time to read this." Yeah, yeah. [laughs] Like, "I'd prefer for you to just tell me."
16:55
Um, and you feel like you may be being really efficient because you're pumping out a bunch of words on, on paper, but, like, does the team really need to know it? Yeah. Yeah.
17:05
And so I think that they've had some pretty honest conversations with me, and it's something I'm trying [laughs] to get better at- Yeah...
17:12
and, and make sure, like, even when I, like, uh, my weekly update at, at the end of a Friday to the team, talking about, like, team wins and, um, metrics that moved or key projects we're working on, I'll say at the beginning of that, like, this email, like, some- Yeah...
17:27
of the below has been written by AI. Yeah. That's right. Just so they kinda know- Yeah... if it's me speaking or if it's- The system- Yeah... or things like that. So- Yeah...
17:36
um, we've also, like, got a special email address now for the system so it knows if it's coming from me or coming from- AI stuff. Yeah, yeah... AI stuff. Yeah.
17:44
How do you look, yeah, at other businesses, um, where they're trying to do similar rollouts but it's starting maybe the bottom up- Yeah... versus at the moment you're going top-down? I think it's...
17:55
This is a personal opinion, but I think it's absolutely essential that all businesses are built with AI in mind or agent-ready or whatever you wanna, whatever words you wanna use to describe it.
18:08
I think, um, you're building for the old world if you're not looking at how- ways you can kind of build some of those systems and automations into the business.
18:18
I think, um, there's al- always fear around that as well, around, like, employment. Yeah. Um, a lot of people think it's gonna take their jobs, and, and one thing I've been really,
18:29
um, open about with my team is, like, it's not gonna take your job. It's just gonna allow you to free up time and work on things that are really important. Going back to your transcripting, like- [laughs]...
18:37
do you still watch two hours of videos to do transcripts- Like-... or- Yeah... you use AI to do that and do something more exciting with your time? Yeah.
18:42
Yeah, like if I'm, if you're spending three hours, like, writing show notes for a podcast, like- It's not-... I'd rather be paying you to do something- Yeah...
18:50
way more beneficial both for yourself and your career- Yeah... and something that's more interesting and, and provides more value to our readers and things as well. So- Yeah... I think, um,
19:01
people thinking about starting a company nowadays very much need to keep those things in mind, choosing software, making sure that they can communicate and connect with these AI tools, and that it's not locking them into, um, any one platform.
19:16
So can you export the data- Yeah... to take it elsewhere? Yeah. Um, can you, uh, have API access to a lot of things?
19:24
Like, um, I pretty much any software I sign up to now, I make sure it has API access so that my- Yeah... my tool can talk to it. Yeah.
19:34
There are ways around it, like we use Playwright- Through the browser or whatever it is. Yeah... yeah, through the browser to automate a bunch of stuff.
19:40
Um, but, like, it's a completely different new way of working, I think, that people need to be leaning into, not just for, like, company insights, but for saving time and automation.
19:51
Like, um, I for- I forget who it was, it might have been Claude or someone else who said, like, there's gonna be a one person, $1 billion business- Right. Exactly... if, if it hasn't already happened. Exactly. Yeah.
20:02
If you were to start a business all over again, like, and didn't have the, the legacy of existing systems and processes, how do you think about, like what is the design of a new business?
20:10
Is it the org chart that changes the most? Is it, like, what are people focused on?
20:14
Like if you- I think, um, we're very lucky in that, and then the other thing I should mention is 'cause we are a smaller business and Frank and I are the cal- co-founders- Yeah...
20:22
we're able to plug and play pretty easily. Yeah. Yeah. Like, we don't have to go to IT for approval. Yeah. [laughs] A larger legacy business has a lot more rigmarole to go through to get those things set up. Yeah.
20:34
So I think it's more around understanding that these things are possible and systems thinking and, and I don't have the answer for a larger organization. Yeah.
20:42
Um, for others, I think, like, starting a business, I think it's going back to fundamentals almost and thinking about, um, like, uh, thinking about how you essentially make money- Mm...
20:59
and what your market offering is and thinking about, not necessarily about, like, dropping AI in. Like, there's the famous story of, um, the sneakers Allbirds who transitioned to become- Right...
21:11
an AI company and then they- Yeah, a data company or AI data or something. Yeah. Yeah, that's right. With... And they didn't actually explain what that actually meant in practice. It's not like plugging in a chatbot.
21:22
It's thinking about, okay, if I have a really good understanding of how I make money in this business- Yeah... how can I chat with AI to make that little part of it much more efficient, much quicker- Yeah...
21:34
much more costly, cost-saving. Um, and I think, I think that's the way people need to think about it. It's not just using AI for AI's sake. It's using it to- Yeah...
21:44
help with particular problems within the business that can, you can set up for the long term. Yeah.
21:49
How do you sort of product manage your own brain if you're the one, if you're the chief engineer and the chief product officer of sort of making sure that you're working on the right stuff?
21:56
Like- Yeah, that's one thing that I, is constant work in progress. Um, I've recently been playing around with Claude's new mods- Oh, yeah... um, to, like, change around my VS Code.
22:09
So on the right-hand side I can see, like, a list of priorities basically. Yeah. So I know what it thinks is a priority for me- Mm-hmm... and I can say, "Okay, that sounds about right. I need to get that thing done.
22:21
That's the most important thing"- Yeah... uh, or the most business critical thing for me to get done at the moment. Yeah. So having that visually is helpful. Yeah.
22:29
Previously, I just used a task manager for that, so I used TickTick. Yeah.
22:33
But I think, um, managing those priorities, using those, like, senior leaders meet, uh, meetings to make sure that everyone's priorities are lined up, everyone's on the same page about what is important, they're not just busy spinning their wheels working on things that aren't gonna move the needle.
22:49
Yeah. Um, I think that's probably helps, gives us a lot of direction. Have I made a lot of mistakes with things like that along the way? Yes, 100%. I used to have, like,
23:01
50 plus different sessions open at a time in Claude Agent view, and I couldn't find where anything was. And, um, I've become much better at making sure I've only got kind of like three or four prompts open at a time.
23:15
Yeah. Other really handy tip is, um, having an audio cue when a, when Claude finishes responding, so- Yeah. Jump back in...
23:23
so it'll ding me, um, so I know what, one of the sessions is done so I can jump back in and chat with it.
23:30
Um, there's a bunch of other things you can do, like- Giving, having like a, an always on context, um, bar at the bottom where it can tell you what that session is about. 'Cause often- Yeah...
23:43
you jump back into a session, you're like, "I don't even know what I was- Yeah... talking about here." What did I ask it to do? Yeah, yeah. What were we working on? Yeah.
23:48
Um, so we do have a huge amount of, like, projects that probably have started, got half done, haven't been finished. I think that's, that's always really difficult.
23:57
But, like, I would say that going back again, back to the fundamentals about how do, what's our business model? How do we make money? Mm. What are those important things that we need to be focused on?
24:08
Um, and always coming back to that rather than getting too bogged down in, like, I want this to look pretty in a certain way, but it's not gonna fundamentally change the business. Yeah.
24:17
Um, or this would be really cool if you automate it or build this dashboard, but is anyone actually gonna use it? Yeah.
24:23
Um, that's kind of where, where I've made mistakes in the past, is like trying to build a chatbot for everything that no one actually ends up using. [laughs] Needing, yeah, exactly. Yeah.
24:32
Is Frank just as passionate about this as you are, or is it something that's uniquely the thing that's driving you at the moment? Yeah, no, Frank, Frank is fully set up, so- Oh, he's, he's vibe coding as well, or?
24:42
Yeah, yeah, yeah. Okay. So he built like a great retention engine, um, for us with our clients. So- Yeah...
24:49
it alerts us now, like, this brand spent with you 12 months ago, it's probably a good time to re-engage with them. Yep.
24:55
Or like, it's been X amount of weeks until you've, since you've spoken with- If you've got a, if you've got a medium, like similar sized business, you know, similar realm, um, but you know, the, the, the founder or leadership is still tied up in day-to-day, like they're still accountable for something else, who is that next best person to sort of own this, do you think, to, to get it up and running?
25:14
Is it someone from a product perspective? Is it a, you know, head of AI? Is it a, how, like how do you think of the org chart?
25:22
Um, it can be really daunting when you look at like setting up some of these processes, but the great thing is there's a huge amount of free knowledge available online.
25:31
If you look up like AI operating system on YouTube, there's a bunch of like two or three hour- Mm... or four hour guides about setting one up for your business.
25:40
Um, and that, that's like, you know, half a day's work to get that done- Yeah... following along with those things. Um, I'd say again, that applies for small and medium sized businesses.
25:52
If the risk is you set it up one weekend and then you never get back to it, like you need the resourcing, like your own resourcing, right, to be able to dedicate- Yeah...
25:59
the time to- I would, I would say that people, like setting up the infrastructure's probably the first part of it. So understanding like how you're gonna set up your virtual private server. Mm.
26:10
Making sure it's secure, like- Yeah... you've, you've set it up in, on a private- Yeah... GitHub repo and- And there's probably contractors and stuff you can easily help that- Yeah, yeah... place.
26:17
That's, that's- Setting up the infrastructure-... good for everyone... is really important. Yeah. Then the, the next step about how you actually make use from it- Yeah...
26:24
is probably the people working in the business, 'cause they- Yeah... should understand the systems and processes the best. Yeah, that they own. Yeah.
26:32
Um, and even like before getting stuck into the automation or the AI, like even going through that process of let's think about the steps within the business- Mm-hmm...
26:41
um, and the SOPs and, and, and systems and how things operate. Writing that down and documenting that, um, will help you massively. 'Cause- Yeah...
26:51
one, just getting an understanding of, oh, this is actually something we could probably automate, and two, once you've got that documentation, you can ha- Half the job is done already to give AI- You can hand it, hand it back to the AI and- Yeah...
27:03
it's got a much better understanding of your business before you even start to automate. Yeah. There's obviously a lot of challenges around it, but- Yeah...
27:10
from an optimism perspective, yeah, what's, what's exciting you at the moment, getting you out of bed? I think, um, for us it's around thinking of ways AI can actually improve the quality of work we're doing.
27:22
There's a lot of stuff around like AI slop and being really generic. Mm-hmm. We've discussed like Claude Design and everything. Yes. Like, you can spot Claude code and, or Claude Design.
27:33
I think for us, we wanna make sure that we're using AI to actually improve the quality of things and, and being able to output really great work in line with our brand, brand guidelines and those sorts of things. So, um,
27:49
that is really exciting to me because I think we're able to deliver much more value to our readers. Yeah. Um, so like
27:58
giving information in better, more intuitive charts where, um, like building a graph or something like that back in the day might have taken you half a day. Yeah.
28:07
Like design a- We could have had, had an engineer for a week to build a thing or- Yeah, yeah... something similar to work.
28:10
So building like an interactive chart for an article now we can whip up in line with our guidelines- Yeah... using AI in space of like five minutes.
28:19
And that sort of stuff is really exciting for us as a media company 'cause we're able to produce, edit, and output a lot more content- Yeah... um, with a smaller team.
28:28
So I think, um, it can be really overwhelming as well that there's expectation that you just produce more with less people. 400 articles a day. Yeah, yeah. And I don't think we're going down that path.
28:39
For us it's more about like improving the quality of what we're, we're currently doing. Wonderful. Scott, it's been great chatting. Yeah. Thanks for joining me today.
