The Launch Episode - Pragmatic AI

Episode transcript

Amelia (00:02.434)
Hey there, and welcome to our brand new podcast. I'm Amelia.

Paul Wood (00:07.293)
I'm Paul.

Patrick Cunningham (00:10.702)
I'm Pat.

Russell (00:13.09)
I'm Russell.

Amelia (00:14.766)
And this is Fin the Week. Everyone, welcome. How are we doing?

Paul Wood (00:20.936)
All good, thank you. Yeah, pretty excited.

Russell (00:25.313)
Yeah.

Amelia (00:25.368)
So.

Patrick Cunningham (00:25.921)
Yeah, super excited to actually be launching our own podcast.

Amelia (00:31.79)
Yeah, we've been talking about it for a little bit.

Russell (00:31.874)
Yeah, it isn't something I thought we'd be doing. But I'm up for giving it a go and seeing what happens for sure.

Amelia (00:44.448)
And, you know, it is episode one, and I guess we should probably start by explaining what we're doing here. Paul, this was kind of your idea, right? So should we start with you?

Paul Wood (00:45.087)
Yeah.

Paul Wood (00:53.984)
Yeah, I mean, so this is an idea we've been kicking around for quite a long time, actually. I can remember a couple of years ago that the three of us sat down and spoke about it. We're always discussing stuff in our industry. We're quite thoughtful people, I think. I mean, others can make their own judgments, but we've always got something to say about what we do. And we said how cool it would be to just sit down on a weekly basis and record these sorts of discussions.

But you know, we never really got around to doing it. So now feels like the right time. And it kind of coincides with we've been thinking a lot about our brand as a company. We're doing a bit of a relaunch in a way. We've rebranded, and we're expanding what we do as a company. We've got a real sharp focus on the industry that we work in. And so launching this sort of thing feels like a great way to share what we do and actually just exercise our own muscles in a way of what we think about, the work we do, how we think about it, all of that sort of stuff. So yeah, that's where the idea came from.

Amelia (02:10.062)
Who do you kind of envisage this show being for? Who's going to be listening?

Paul Wood (02:14.986)
So our kind of focus as a company is on working with firms in the financial services sector. Our background is that we started up in Guernsey in the Channel Islands, which is where Pat and Russell are both sitting. And so Guernsey is known for being a kind of finance hub of the world. So we naturally got exposed to that world. And then over the years, we've focused in on that.

So the work we do and the things that we tackle day in, day out are of interest to people in the financial services industry, particularly marketing teams and operations teams, because they're the people we work with every day and they come to us with the challenges that they're facing. We learn a lot about what they do. We know a lot of stuff about the industry, but we also know how technology integrates with that part of the industry. We'll probably get into discussions about regulatory considerations, which is often seen as a bit of a pain to talk about, but actually is quite interesting. It gives you a really kind of rigid box to work within that makes you have to be creative in what you do.

Amelia (03:33.524)
And you know, these are the sort of people working in that sector who will be listening. When do you think these people will be listening? What do you kind of envisage when our listeners are listening to the podcast?

Paul Wood (03:44.746)
So I listen to a fair few podcasts myself and I like to use them as almost a part of my weekly routine. Like I'm into guitars, for instance, and I listen to a guitar podcast that comes out every Friday lunchtime just because it's a part of my weekly routine. So my ambition is that this will become a part of our listeners' weekly routine. It's not going to be dry or stuffy. We're hoping that it's just capturing the discussions that we have all the time and the kind of messy thinking that sometimes goes on, and so it becomes like a light-hearted part of the start of every Friday morning, which is when we're going to be launching each episode. So that's the plan. Yeah.

Amelia (04:31.416)
So people can be having it on in the office whilst they're working on a Friday morning, kind of rounding off the week.

Paul Wood (04:36.456)
Yeah, yeah, exactly.

Amelia (04:38.926)
And what we're thinking we will be covering stuff that's kind of topical or more of those core concepts?

Paul Wood (04:46.794)
So I think and the other guys can probably have their own view but I think a bit of both, to be honest. Something else I should mention is that with the launch of our new website, we're taking a very sort of publication-first approach. So the website will launch a new edition every quarter. So each quarter there will be an edition. We'll get into the theme of this quarter's edition shortly.

The idea will be that we'll have a kind of pre-planned programme of things to discuss, but because our industry (it's a bit of a cliche) is constantly evolving, it's constantly changing, particularly at the moment with AI, which is what we'll be talking about today. And so I dare say we'll have to sort of programme in, you know, emergency episodes when something big happens, because it does happen.

I mean, I know, Pat, you're at the sharp end of AI in the development world at the moment. And the pace of change there is insane, isn't it?

Patrick Cunningham (05:55.693)
We could do a daily emergency podcast at the moment. That's what's going on.

Paul Wood (05:58.4)
Mmm. Yeah, "emergency" is probably the word. Yes.

Amelia (06:02.082)
Pat's going to be in charge of the breaking news.

Patrick Cunningham (06:05.177)
Yeah, just put on a permanent live feed. 8am to 6pm.

Paul Wood (06:12.8)
Strap a GoPro to your head.

Amelia (06:17.366)
So that is kind of what it's all going to be about in a nutshell. And I'm sure, you know, as we go through over the coming weeks and months, it might evolve, but that's kind of the concept. And usually this part of the show will be about tackling the big talking points of the week. But this week, why don't we talk about the theme of that first edition that you mentioned, Paul, which is pragmatic AI at the risk of triggering those

Paul Wood (06:18.325)
Yeah.

Amelia (06:41.352)
sick of hearing about AI. Why pragmatic AI, and what do you mean by that term?

Paul Wood (06:48.064)
Yeah, so when I was mapping out the programme for the first few quarters, it felt obvious that we had to talk about AI because it's just so critical to I mean, everybody is talking about it in one way or another, certainly from what I can see. But it also struck me that with anything like this that grows so quickly and so big, there's a lot of hype and a lot of stuff that isn't necessarily true, or a lot of people worrying about things that aren't necessarily worth worrying about. But I also think that because the work we do is heavily impacted by AI and we're all just genuinely interested in it, it's easy to forget that there are people in the real world who, you know, actually just dabble with it or don't really know that much about it at all.

Something that made me chuckle the other day actually was that Pat and I took part in a webinar a couple of weeks ago talking about AI and its impact on the communications industry. And my wife happened to join and watch it. So I jokingly said later on that day, "What was your main takeaway from the webinar?" And her response was, "I still don't know what an LLM is." She was embarrassed because I laughed about that, but actually it was a good reminder that, you know, normal people don't really know anything about this, and they don't necessarily care all that much until something lands in front of them that is really obviously helpful.

Coming back to the question pragmatic AI what we found is that behind all of the hype, there is some genuinely really useful stuff you can do, which sounds a stupid thing to say. But even in the last couple of months, the amount of change it's had on production I mean, Pat, you mentioned a little while ago something that kind of was a light bulb moment for me, which was that AI in development isn't about creating sort of your own AI chatbots or anything, which is what a lot of people have focused on to this point. It's actually just

Paul Wood (09:10.579)
changing the way traditional software development happens, and it's making it it's democratising it, basically. It's making it so much easier. And I mean, Pat, you can probably expand on that.

Patrick Cunningham (09:21.379)
Yeah, I mean, I think I'll start by just giving my view on sort of pragmatic AI. I think it's quite easy to fall into one of two camps with AI. There's the sort of fear camp and there's the hype camp. The fear camp is "I'm worried about data protection, I'm worried about bias, I'm worried about hallucinations and inaccuracies. So I'm not going to touch it. I'm not going to let it into our business. We're just going to be super cautious." And the hype camp is "It's going to 10x everyone's productivity and everyone's going to lose their jobs, or, you know, we're going to be able to triple our sales overnight." And neither of those extremes are correct, really.

So pragmatic AI is sort of trying to straddle both of those and find a sort of happy medium in the middle, where actually you're using AI securely and in a useful way that's improving your business without causing lots of fear, and whilst also improving your productivity but probably not 10x-ing it just yet.

On the development side, to your point, Paul what I've been talking about there is the fact that I've kind of had this realisation over the last couple of months that the biggest change that's going to come from AI, at least in the next year or two, is not through AI agents replacing people and AI automations automating huge swathes of the economy. It's going to be from developers using AI to build much better software. And what that means is that software will work more smoothly. It will integrate better. You'll have interfaces that become much easier to use, that are more accessible.

Paul Wood (11:16.511)
Mmm.

Patrick Cunningham (11:19.545)
And so I think over the next couple of years, we'll suddenly see you know, the world is full of technology, but most of it doesn't work very well. If you've ever tried to kind of get Echo or Amazon Alexa to do something beyond setting a timer, it just generally fails. Siri is not that useful at all; it's a bit further along. But we'll see things like that just move much, much more quickly because as a developer using these AI tools, you can get so much more done to such a high level of quality in much less time.

So yeah, we'll see just the general progress of software and technology start to accelerate massively. And off the back of that acceleration will come more powerful AI tools, and AI will become more integrated into our day-to-day lives. So yeah, that's sort of what's happening on the development front for sure.

Amelia (12:19.01)
I think that's quite interesting. Sorry, I was going to say I think it's quite interesting what you were saying about this kind of fear/hype thing. Because Paul, I'm probably, you know, like your wife I don't really use AI in my day-to-day work. I use ChatGPT occasionally, but not really. And for me, most of the conversations that I see around it are very much around risk, danger, ethical issues. And do you think that, you know, as obviously people who know a lot more about it, that maybe means that everyone's just completely missing the point because it's overshadowing the conversations about its real potential?

Paul Wood (12:19.357)
Yes, that was good.

Patrick Cunningham (12:50.573)
Yeah, I mean, there are data protection issues with using ChatGPT. You can't get around that. But honestly, if you have a conversation with ChatGPT about what you're going to have for dinner tonight, or you share some information about a symptom you're experiencing and it comes back with some advice, that data isn't really going to be leaked anywhere. That's low-risk stuff. If you're using ChatGPT to crunch through a thousand CVs for a job application and you're using it to filter out relevant applicants that's the high-risk area where bias is probably going to come into play.

So it's about how you use it and what you use it for. If you're worried about bias, there are some very clear-cut areas where you shouldn't be using large language models, but there are also loads of areas where you can use large language models without worrying about bias. For example, processing invoices or crunching numbers, you know, writing code, that kind of stuff. If you're worried about data protection, there are ways you can use AI safely. You can run models within your own Azure cloud. You can use models on Amazon that are within the EU so that they're GDPR compliant. And technically there are ways to kind of get over those data protection concerns.

And the other common issue is hallucinations and that, you know, large language models do hallucinate a lot, but again you can really mitigate those risks by prompting well, using the right tools and sort of structuring your data in a specific way in order to mitigate those risks and produce an output that's incredibly useful. So yeah, it really comes down to kind of how you roll it out and what you do with it and the performance that you use. So yeah, that's kind of addressing the fear side of it.

Paul Wood (14:52.029)
You mentioned I was going to say, you mentioned hallucinations there. I'd be interested to know is that a common term? Like, do people know what that means? Amelia, does that mean anything to you as a non-techie?

Amelia (15:06.87)
No, no, it doesn't.

Patrick Cunningham (15:09.325)
Yeah, I guess you mention hallucinations to people and people think you're either on drugs or having an episode. These days, in the context of AI, hallucinations, it's a term that is used to describe when a large language model like ChatGPT confidently gives you an answer that is completely incorrect.

Amelia (15:13.23)
You.

Patrick Cunningham (15:38.457)
So an example of where hallucinations have caught people out in recent times there's a famous story of a lawyer who was representing someone in court who got injured in a plane by one of the trolleys; it was rolling back and forth. And he asked ChatGPT to research into a series of precedents for this case so he could present some precedents to the court.

ChatGPT supposedly researched into them and hallucinated a dozen cases a dozen similar cases and similar outcomes which he then put straight into his court documents. But they were all made up. ChatGPT just made them up out of nowhere and confidently said these are real. And he got in a lot of trouble. And it's happened a couple of times since then as well, with lawyers asking ChatGPT to do research and it's hallucinated precedents.

And it does get things wrong, quite a lot. But there are areas where it will get things wrong and there are areas where it won't get things wrong. So if you give it a long document and ask it to summarise that document, it will rarely make things up about that document because all the information is there. You're not asking it to plug gaps, if you like. And so it will generally get that summarisation really, like, you know, correct. And it will talk about the document in an accurate way.

If you ask it to research into a load of precedents without actually going into depth, then it might make them up. But if you flick on a deep research mode, which is a mode available in most kind of online chatbot tools, it will go away and it will research the internet and it will take five to ten minutes, and it will think a lot more deeply about it, and it will critique itself and review all of its citations. And actually, if you flick on deep research mode, it will generally be completely accurate, and it will also give you references and citations that you can follow up on and check. So you know, there are ways around the hallucinations, so you definitely shouldn't be concerned about that instead you should be researching into how you can address them and understanding when they happen and what can trigger them.

Amelia (18:05.314)
So just like with any tool, you've just got to know how to use it properly, basically.

Patrick Cunningham (18:09.675)
Exactly, yeah. And I like to use the phrase: don't caution yourself out of existence. In the context of Guernsey, in the finance industry here in particular, they're very risk averse. There are a lot of compliance rules and regulations they need to adhere to. And as a result, they're very cautious about rolling out new technology in case it results in them falling foul of compliance regulations.

But honestly, right now, there are a few firms in Guernsey that just have a blanket ban on AI. That's definitely the wrong move because they're going to fall behind due to an abundance of caution. And fundamentally, they might not be able to catch up if they fall too far behind, because the world is just moving so quickly. So yeah, I say to people: don't caution yourself out of existence. Experiment and be pragmatic about it.

Amelia (19:14.382)
And I feel like let's get a little bit more specific. So have we got any real examples of where AI is actually changing things currently?

Paul Wood (19:25.215)
I was going to say, we've done a lot of prototyping recently. So Russ, you've been involved in a number of things. You've built your own concept for an app, and it's possible to prototype now, isn't it, in ways that just wouldn't be possible even maybe six months ago.

Russell (19:49.056)
Yeah, I think in that respect, it's a really exciting time for people who have an idea for an app. I think in the past in fact, the app that I produced a prototype of, I think it was sitting in a business plan for probably six or seven years. And then within a week using Figma Make, I had a working app which basically tested all of the ideas that I had in this business plan. I probably built it in a few days. I produced a demo and put that online. You can see it on my LinkedIn. Shameless LinkedIn plug there.

But I think in the past, it probably would have sat in that business plan perhaps for another few years until I convinced myself to spend six months writing the front-end code, or

Paul Wood (20:28.456)
You.

Russell (20:43.65)
persuaded a developer to come on board and create an initial prototype. I think what AI allowed me to do is essentially go into the design software that I'm already using, take my design concepts and bring them to life almost instantly. I think across about 200 or so prompts, I produced a prototype which had a landing page, had system settings, had all of the elements which allowed me to now demo it to a potential customer visualised that business plan almost instantly to see if it works or not.

I think it was still a great exercise not only to see how quickly you can use AI to kind of put together a prototype, but also to fail quickly if it does fail. And I think that that's what we're going to see more now. We're going to see entrepreneurs kind of failing faster, building these prototypes and understanding if their concepts work more than ever before. And hopefully as a result, alongside all of the kind of I don't really like the phrase AI slop which is appearing, I think we'll probably find that those good ideas can succeed faster just due to the rapid pace that AI gives us in producing prototypes.

Amelia (22:23.854)
Can I ask, Russ how much time would that have saved you? So, you know, if we go back, you said, Paul, like, you know, six months ago even, but if we went back a year or two, if you were to do something like that, how much time would that actually take you?

Paul Wood (22:25.043)
Yeah.

Russell (22:37.41)
Well, I mean, front-end coding isn't my speciality, although I can do it. I mean, if I sat down and wrote that much code, I'd probably be doing it full-time for like a few weeks. And I'd probably produce the whole concept using prompts in a couple of days. Yeah.

Paul Wood (22:58.399)
Yeah, the barrier to entry is just so much lower. And I think that's a good thing because it means that people are not afraid to try ideas they have. And like you say, Russ, if it fails, it fails quickly. Whereas a year ago, your concept was sat in a business plan and realistically wouldn't have got built in any way because no one would have had the guts to sort of back it and put the budget behind it for long enough because you're just not sure. Whereas now we can all test this prototype. We've got some conversations lined up with people to show them, and we'll know quickly if it's a go or not.

Patrick Cunningham (23:45.603)
Yeah. And then when it comes to if it is a goer and when it comes to actually pushing the button and building it into a production-grade application, then that whole process is going to be a lot easier too. You still need to be an experienced developer in order to take one of these sort of prototypes and build it into a robust scalable application, but a developer armed with a powerful AI tool like Claude Code or Codex can really rapidly get something decent together in a fraction of the time.

And what we're seeing on the development front is we're saving a lot of time and building stuff much more quickly, but it's allowing us to spend more time on refining and polishing that end product. We are seeing significant time saving, but we're using the time saved to really, really elevate the quality of the work we're doing. So we're producing software that's much more stable, that's got much more automated testing built into it, so there are fewer bugs, it's more accessible, it's more feature-rich. We're able to build much more complex and sophisticated features than we would have done in the past.

And I say to people that traditionally software development has always just been a journey of compromise, because it has been so time-consuming in the past to build software. And you've always got a limit on time and a limit on budget. And so as a developer, I don't think I've ever been involved in a project where I've been completely happy with what we've produced. Because as a developer, I'd love to spend three weeks working on this specific feature and polishing it and making it perfect and completely bug-free. But in the real world, you've only got two days to build that feature. So you make it good enough to go into production. But

Patrick Cunningham (26:09.497)
often you have to say to the client, "Well, we can build this feature, but it might not be able to do this thing, or we might have to pause this thing until later, put it into phase two." So there's this constant negotiation between how much time and effort we've got to do something and how much it would actually take to do it properly or to do it to a very high level of quality.

And what we're seeing now is that, as a developer, a lot of those kind of boundaries are being removed. All those roadblocks are being removed, so you can really go to town and build something that you can genuinely be really proud of in terms of the quality of that end result. And of course that's allowing us to deliver a lot more value to our clients as well. So actually it's really helping our agency currently, because we're able to do work on budget. We're able to do work to a much higher level of quality. There are way fewer difficult conversations with clients about compromise, because the conversations now are more along the lines of: the client asks if we can do something and we just say yes. No problem. You know, more often than not, it's like, "Yeah, we can do it in the budget as well. We don't have to charge you extra for that." So everyone's happy. And it's just at the current point in time, it's just a really great technology for everyone. But yeah, who knows what it's going to look like in two, three, four years? You just don't know.

Paul Wood (27:48.115)
I think one of the big fears is that it's going to destroy jobs, isn't it? That, you know, if you're able to do so much more with the same team, then are there going to be more jobs available?

But I watched an interesting report from Sky News the other day, which was talking about how they referenced (I'm going to butcher the story) but they referenced when steam production machines were created, it made production so much more efficient that people involved in the coal industry were worried that demand for coal would just drop off a cliff because you just need less of it. But actually it went the other way and it rose exponentially because people wanted more of a good thing. So they wanted to use these machines to do more.

And actually they looked into the data for the jobs market at the moment, and demand for developers so the examples we've been talking about here has actually gone up recently, because I think people realise you can do more with those people, and so they want more of that. So it'll be interesting to see if it does go that way. Because yeah, if the barrier to entry for creating new things is lower, then I guess people are just going to be going for it and creating everything they can. Companies are going to be creating their own software, whereas in the past they might have used a SaaS piece of software. So yeah, could be really interesting. It might not be the disaster that everybody's predicting for the jobs market. I don't know.

Patrick Cunningham (29:34.659)
Yeah. And on that effect, I think there's this rise of people just going into tools like Lovable or Replit and just vibe coding their own products because they've got an idea and they can just do that in an afternoon or over a few days. And then what happens is you've just got all of these thousands and thousands of different products out there. And some of them because they've been able to prototype them some of them have found there is demand for those. And all of a sudden they need to take that vibe-coded prototype and turn it into a robust, production-grade application. And that's when they need developers to come in. So there's this emerging market of demand for developers who can take a prototype and actually turn it into a secure production-grade app. That's really interesting.

And I think the SaaS world is definitely kind of ripe for massive disruption as well. You mentioned there, Paul, that we'll probably see firms building their own software, which is definitely true. And I think there are a lot of SaaS products out there that kind of help to organise knowledge platforms like Salesforce, monday.com, Jira, you know that potentially can get quite expensive too. It's not going to be that hard. In fact, it's not that hard right now to build your own version of that software that's aligned with your own company's specific requirements. It's cheaper to run. It's much more lightweight. It's much more aligned with how your business works. It's not bloated. We're definitely going to see a rise in that trend. And so I think we'll see these huge multinational SaaS firms probably start to lose revenue and probably change their business model. And we'll probably see a rise in firms building out their own bespoke software like small firms.

We've done it. We recently swapped out our ticketing system Zendesk for an open source platform that we extended significantly using

Paul Wood (31:51.935)
You.

Patrick Cunningham (31:57.217)
AI technology to align with our requirements, and it saves us an absolute fortune. You're going to see that across a lot of businesses, I think, because from a budgetary perspective, it's feasible to build your own software in a robust way and actually save money over paying monthly fees to Salesforce or Zendesk or whatever.

Paul Wood (32:20.871)
It's a better experience as well, isn't it? Because some of the most woeful applications out there are those major tech platforms. You know, we're yeah. I mean, yeah, I despise having to use Jira and things like that. It's just so difficult. And I'm sure there's someone out there it's perfect for, but

Patrick Cunningham (32:30.669)
Yeah, they're definitely complacent. Definitely complacency there, yeah.

Patrick Cunningham (32:48.419)
Yeah.

Paul Wood (32:48.66)
you know, having your own bespoke platform is just so much nicer.

Russell (32:52.162)
It kind of opens the door for three things. It allows agencies like us to build a whole suite of proprietary software to run our agency, because we have the capabilities to do it in-house. It allows us to create apps which we can license to compete with those large SaaS products as well. And also, just outside of the whole SaaS market, it just allows us to create enterprise applications at a much larger scale with a smaller team and reduce the cost of those.

I think I heard someone talking the other day that now is really the only time where a small agency of ten people could potentially be billing a billion-pound revenue or be like worth a billion pounds because the output now is so much greater than it was before that a team of ten would just really struggle to compete with these big agencies of, you know, of thousands of people. Whereas now we're in a position perhaps where we can.

I think the only kind of risks I see is that the landscape is going to change quite quickly over the next few years, and having a competitive edge in all of this will eventually fade and probably merge into one kind of industry-standard way of running an agency in the next few years. So you have to keep on evolving and staying ahead. And I think right now as an agency at Indulge, I think we're on the front foot there

Russell (34:37.93)
with what we're kind of experimenting with and what we're adopting into our processes.

Paul Wood (34:49.183)
I was chatting to someone literally before we came to record this session and we were talking about AI inevitably, and they mentioned that it feels like the value that anyone can bring, certainly in the agency world, is judgment. That's always been a part of the value having good judgment, making good decisions, creating the right strategic direction.

But then the other big part was the production and the kind of creating things. But increasingly the production side, certainly the nitty-gritty part of it, is being handled by or at least augmented by AI. But the thing that hasn't changed and I don't think will change, certainly in the short term, is that judgment and that ability to look at a solution and say, "Does this fulfil the requirements?"

And as Pat was talking about, the polish and the kind of perfectionism element of our jobs it allows us to then do that better. So yeah, it's going to be interesting to watch how a lot of organisations sort of shift focus, I suppose, into that judgment area and looking at making decisions, but leaving the hard work the coding, the creation to bots.

Russell (36:13.698)
And also I think the human skills are going to be more valued, and I think you only really gain those human skills by well, at an agency, I think some people just naturally have them. But I think ultimately people want to work with us and want to work with agencies as well as the services that they provide just because they have a good relationship with the people who are providing those services. That kind of growing relationship and understanding our clients, being on that longer journey with them to improve their business, is something that AI won't replace, because we will still be accountable for the results which AI gives them. If we're speeding up production, someone's always accountable for what AI produces along the line, and ultimately it's the people producing it.

So, you know, I think it doesn't take that away. If something isn't working, you can't say, you know, "I'll just ask the AI" or, you know, "let's have a look what the AI has done and try and figure out from there." I think you, as a producer using AI, you're always going to be directing AI as a production tool to deliver what you intended it to. And I think that that's really important, because if you allow AI to kind of make up the interface or build the app in its own way, then how can you defend that and become accountable for that if you're having a conversation with the client? So I think that's really important when you're using this technology. And that comes back to, you know, humans being responsible for the strategy, being responsible for

Russell (38:01.22)
testing and validating like we were talking about, and that's going to be the key element, I think, in rolling this out successfully.

Patrick Cunningham (38:15.17)
Yeah, definitely. I think in the context of an agency this applies to everyone but you should use AI to address your weaknesses and not replace your strengths. And the strength of an agency is its people, and the fact that an agency has a team of experts who can achieve a much better end result in a particular project than the client trying to do that project on their own. And an agency using AI, with people driving that AI and using that human judgment to produce an incredible end result, will always produce a better result than the client attempting to do the same thing using AI.

So the value proposition of an agency is that the people in the agency can produce a better result than the client can produce. There's an argument to say that if the client starts using AI, it can close that gap to what the agency can use. If the agency is also using AI, you're maintaining that value proposition. And that will remain. I genuinely think that people will continue hiring agencies because the client will have more faith in the agency delivering a successful result using AI than themselves delivering a successful result using AI.

And like you say, Russ, as an agency owner or a production person in an agency, you definitely shouldn't have AI doing everything, because then you lose that value proposition. You should have AI sort of crunch code and kind of close the gaps on certain parts of the project, but your human judgment and your professional experience is really where the value is. And so you need to retain that, because that's what is going to make your product stand out and really solve the challenges that the client has engaged you to solve.

Amelia (40:17.486)
I think that's maybe a place to leave it for now. But I know we're going to be exploring a lot more about AI in the coming episodes, so plenty to sink our teeth into. So now I'm looking forward to this. This is a feature we're going to be doing every week. It's called Jargon Busters. So basically we're going to see how well you guys know some of these kind of industry terms. How are you feeling about this?

Paul Wood (40:42.749)
I'm kind of not confident, but I'm looking forward to it, I think. Yeah.

Amelia (40:42.798)
Are we feeling confident?

Russell (40:42.84)
A bit nervous.

Patrick Cunningham (40:45.112)
Yeah, we'll see. I've got Claude lined up ready to help me answer these questions.

Paul Wood (40:52.989)
Yeah.

Amelia (40:53.865)
So I've got this Jargon Buster list, and in every episode we are going to well, you guys are going to give it a go to see if you know what it means. So the first one is the term "unicorn".

Paul Wood (41:10.427)
Hmm. I'll see. I think I know this. Yeah, this. Yeah, my daughter. My

Amelia (41:11.726)
Does anyone have any clue?

Russell (41:14.132)
Yeah.

Patrick Cunningham (41:15.192)
It's a horse with a horn and rainbows coming out of its bum, right? I know what a unicorn is.

Amelia (41:16.855)
Hahaha.

Paul Wood (41:23.965)
Yeah, my daughter loves them.

Amelia (41:24.216)
Smashed it.

Russell (41:25.93)
I think so. Yeah, I was about to I was actually, my daughter's water bottle this morning when I was getting it ready to put in her bag for nursery has unicorns and rainbows over it. So maybe it's best to ask her.

Patrick Cunningham (41:28.876)
Yeah, I love them too, to be fair.

Amelia (41:34.563)
Haha.

Paul Wood (41:42.907)
Yeah, every new toy my daughter gets, I say, "What are you going to call it?" And it's "unicorn sparkle" every single toy.

Amelia (41:51.662)
So apart from the horse with the horn, what do we think?

Russell (41:51.947)
Yeah.

Patrick Cunningham (41:52.92)
Sparkle!

Paul Wood (42:00.949)
A unicorn is isn't it a sort of shorthand way of describing a high-value company? Like a sort of, you know, these sort of companies that come out of nowhere. OpenAI is a unicorn, I guess, because it sort of came out of nowhere, now it's worth billions. That's the way I understand it.

Patrick Cunningham (42:22.824)
Yeah, like family-y might employ like 30 staff that sort of was launched like two years ago and it's suddenly worth like a billion. It's like an incredible investment opportunity, yeah.

Amelia (42:31.202)
Yeah, you're on the right yeah, exactly.

Russell (42:33.794)
Yeah, I think I'm kind of in the same boat. Like, because it's quite it's almost, yeah, like you say, it's that one-in-a-million kind of company which has done well, and that kind of is the unicorn, because it's special, I think. Is that right?

Amelia (42:55.52)
Yeah, well, basically I mean, it's specific. So it was a private startup company, but valued at over $1 billion.

Russell (43:01.271)
Right.

Paul Wood (43:02.791)
Mmm. Yeah.

Patrick Cunningham (43:04.184)
A billion dollar right? Yeah, I think there's something on these lines.

Russell (43:04.618)
It's specific. Yeah, private startup. I didn't realise there was such a specific explanation for it, but yeah, I think we were on the right track.

Amelia (43:13.218)
Yeah. I like the idea of this super sweet term for, you know, this big finance term. But yeah, you did pretty well there. As I say, plenty more chances with this because we're going to do this in the coming weeks. So probably get to start revising. But for now, let's talk about next week.

Amelia (43:36.142)
So a little preview of what we're going to be talking about next time and this is a big statement. So the episode is titled "I'm going to replace my design team with AI unless you tell me not to". Russ, are we going to replace you with a robot?

Russell (43:51.522)
Well, I think it really I mean, design is so broad. You know, I'm assuming we're talking in the context of technology and probably website and app design, because that's our kind of bread and butter. But yeah, I think the word "design" is quite broad. You know, design's been around for thousands of years. The pyramids were designed. The chair that you're sat on is designed. Websites are designed. And I think ultimately, I think there are aspects definitely and we've already been talking about them web design and app design, which are going to be sped up using AI. I think the methodology behind what makes good design and how you get there remains the same from all the things I was talking about.

What would actually be quite interesting is just throwing that back perhaps to Pat and Paul what do you think design is? You know, what do you think, what is the meaning of design as a word? What is design? And I think that's perhaps something to think about before the next call as well, because I think understanding the meaning of that word and what it is, I think will help us answer that on the next one.

Paul Wood (45:04.849)
Hmm.

Paul Wood (45:18.577)
Yeah, yeah, that's a point. I didn't even think about the chair I'm sat on. Yeah, yeah, no, I'm doing this weekend.

Patrick Cunningham (45:23.064)
Quite philosophical, isn't it?

Amelia (45:23.886)
You've got some homework there, guys. So yeah, looking forward to kind of getting stuck into that a little bit more next week. But for now, that's episode one done.

Paul Wood (45:37.137)
Yeah, exciting.

Amelia (45:37.934)
Very excited, we've started. So yeah thank you so much for listening, and we will see you next week.

Russell (45:40.748)
Yeah, we did it.

Patrick Cunningham (45:41.154)
Ready for the big time.

Paul Wood (45:41.245)
Yeah.

Paul Wood (45:49.479)
Yeah.

Patrick Cunningham (45:49.752)
Thanks, everyone. Cheers. Bye.

Russell (45:50.38)
See you later. Bye.