Tom: I'm Tom and this is John. parts ⁓ and two, we spoke about our general findings and comparisons across ⁓ and Australia infrastructure. ⁓ In part two, we discussed around putting and data in place ⁓ and transforming as part of ⁓ improving the performance of infrastructure. John, what are we to talk about in this?
Tash Wilson: You've been listening to Transform by GHD. Big ideas, real impact. Welcome to Transform, tomorrow's thinking today. Real conversations on the innovation shaping businesses and our society.
Tom: heart.
Jon: I think part three to conclude, what happens next? know we've both in the UK and in Australia, a to varying degrees ⁓ data platform landscape. We've ⁓ got great stuff in some areas, some challenges. sometimes in the same areas, sometimes in other areas. got this utopia of, you know, was the AI revolution has, the phase of it ⁓ has ⁓ software, applications, platform piece becomes trivially easy to solve. got a lot of and hopefully a lot acceleration and reduced costs in that space. How do we get from where we are to actually doing that for some organizations that AI revolution is going to happen incredibly fast. Those organizations probably aren't managing safety critical infrastructure. ⁓ And we a bit about how ⁓ kind of not really possible for lot of the that our clients to move that fast.
Tom: we get there. Yeah.
Jon: But we still want to be part of that. We still need to adopt those technologies. We still want to do things that are going to take us on that journey as quickly as possible. What are those things? I think it's a really interesting question. ⁓ And a very tricky executive question in terms of how do we plan that journey in that's never been done before and is completely new for everyone. ⁓
Tom: No, no, it's not out of the book. Yeah.
Jon: There's no off the shelf plan here. do we plan out a sequence of activities and interdependent work streams that's relatively regrets what our timelines are in that. that's conversation for this morning. in terms of that planning piece, We talked a bit about making the business case for and getting that lined up. do you make the business case for a journey that is inevitable? ⁓ everyone's seen AI, everyone's, ⁓ know, you chat to you, GBT the phone. We know this is coming, but got a completely intangible value, there's no fixed timeline. what's that conversation?
Tom: Yeah. Yeah, yeah. Well, and also before the business case there's almost some managing expectations. ⁓ So not like looking at the tech industry to say, yeah, why do we not have adoption now? It a three to five journey. ⁓ But we as an need to be comfortable with that given the magnitude of the challenge, but then also the benefits of getting it right. ⁓ And then it's not kind of whizzy
Jon: Yep.
Tom: case studies and good things to talk about our conferences, although that is always good. But it's the kind of a ⁓ a build, a build, a build to ⁓ release some quite big benefits of which some of your investment across clients has already taken place. In the UK, we've seen that we worked in organisations where millions and millions was being pumped in for large systems and software where
Jon: Yeah.
Tom: benefits of those still haven't really been realized.
Jon: Yeah, absolutely. there's real kind everything's different, some but everything's somehow still the same. We're going to get that journey right ⁓ by connecting our efforts and ⁓ what we're doing to use cases deliver bottom line that deliver clear pound signs, clear dollar signs. And that sort of first principles always been true. Like we do our best work in the data and technology as a sector ⁓ when it's connected to it's not just gazing and tidying things up for the sake of it, it's connected to the ⁓ business case, the frontline value. So I think of those first principles capabilities are still going to be required to steward that journey. ⁓ But now the technology set the tool set and kind of the awareness of the horizon needs to change. And it's no longer about SQL and Python. It's about kind of understanding what the timeline is for the next large language model release and being able understand for those large language what things that we can do
Tom: Yeah, yeah, yeah, yeah,
Jon: to constrain them, to guide them, to harness them and focus them to ensure that they are delivering correct insights for our specific situation. How do teach them? How do you train them?
Tom: Yeah, yeah, yeah. And then, I mean, in some ways in this case, AI is our friend because ⁓ kind of sets ⁓ vision of where we need to go because data and systems need to be structured in a certain way for you to be able to make use of AI. And then therefore, that roadmap is set. If you don't do it in that way, then you're not going to see the benefits.
Jon: Yeah, yeah. we've spoken in both episode one and episode two, ⁓ about data quality, but the extension of that in ⁓ a AI world do understand Do we understand our own corporate logic, our tacit knowledge? Have we made our internal processes and structures machine readable in a way that we can input that to that AI workflow. I that there's a key kind of shift in the level of recursion that's now possible. Like it's no longer linear flow of source data goes in an application is transformed.
Tom: Yeah,
Jon: an insight comes out, kind of source data, AI has a go and then iterates and makes itself better 400 million times and then there's an and a very different kind of way of thinking and way of operating in ⁓ a of places.
Tom: Yeah. So what are our initial thoughts on what those steps are? And I guess with a reference to the discussions we had in Australia and the clients we're working with in the UK, some of which have similar systems, but they're regionalized as different instances of the same system. very kind pocketed data sets ⁓ is starting point, isn't it?
Jon: No time ever in planning. no time really wasted in improving your data ⁓ taking to put your data first and connect it to frontline value. So even you're not doing related to AI, efforts and projects and of. expended resource to connect your source data to frontline value, understand data quality, understand where you are and build that value chain. Even using, as I say, tooling, ⁓ still going to be worthwhile because the better that is, the faster you will be able to adopt ⁓ AI ⁓ the end of it. or when the time comes for you to be able to do that and effectively. I think the thing that we saw, we have a couple of clients here in the UK that are ⁓ expending serious on. And there were maybe a couple, maybe one in ⁓ Australia was really ready for this, but it wasn't a huge conversation that we had. How do we make our
Tom: Hmm.
Jon: corporate logic, machine readable, how do we start reaching and really understand the semantic meaning in our data and how our data sets are linked together both within the applications that we have and how do they link to reality and making that whole piece machine readable so it is ready to then constrain and guide large language models as develop insights from the data. give much ⁓ accurate insights and much more effective outputs speed. that's ⁓ kind a emerging conversation that obviously, as we've ⁓ discussed before, is an of the data quality piece. But again, this isn't something that everyone has already done or lots of people have already done. Some people have done it. There's some really interesting work in the space. there's no Wikipedia article for this Right. So there's a lot of exploration to do. But it's exploration that's going to have value because it's ⁓ not huge cost in the grand scheme of It's thinking big thoughts and really, ⁓ you know, getting to grips with what
Tom: Yeah. Yeah, yeah.
Jon: do we really do and as a business and as a set of people and as an organization and how does our data to that. But that's going to have immense value when the time comes for us to start in the AI tools.
Tom: Yeah, yeah, yeah. And then I don't know if we skirted over this a little bit, but to the ⁓ the business case. we've examples in the where organizations have tried to kind of value the data if it's at certain quality that the business case hangs off that. Is that the way to do it? other approaches.
Jon: think that particular case, and it was interesting because that was a client in the UK that had ⁓ a really extensive data valuation. And was a number of clients in Australia that were aware of that and ⁓ had to that or understand how they could do something similar. think what's probably important, the pace change that we're currently seeing, is that
Tom: Yeah.
Jon: the business case simply made out of fear of if we don't try and sprint at this, we're going to get somehow left behind or get heavily disrupted off the of the planet. That's not the right answer. The right answer is.
Tom: you
Jon: ⁓ getting a good understanding of what the future is going to look like, getting a coherent plan and going on the journey because with AI, similarly all the applications and platforms and all the bits of technology that you could buy previous, there's still going to be the pitfall. There's still going to be the opportunity to expend a bunch of effort, do a bunch of stuff not really make any sustainable progress or have a misfire with the business because we've done something but we didn't bring the people along with us. all ⁓ those same pitfalls are still going to be there. So it's to be possible to try too hard.
Tom: Yeah, yeah. ⁓ ⁓ an opportunity here that it's, you don't need to procure a massive multi-year contract with a big single supplier. Meaning that you could have put your eggs and your cost all in one basket that actually you can have that roadmap and that plan, but kind of stage gated. So you're, you know, we're doing this, this is effort that we need to put in. This is the support we need from other organizations.
Jon: Hmm.
Tom: then we'll get to here and then we'll, we think we know the next step but we'll take stock and then we'll recommission from that point.
Jon: that's an area where the more mature clients that we have here the UK that really are focusing on surfacing corporate meaning and putting their data first and their data at the heart of rather than their applications, going to be able to the best use of that. ⁓ data is going to drive and then the application piece becomes a triviality and vendor lock-in could potentially become a thing the past. We didn't talk to anyone that had got to that stage yet ⁓ in Australia. And I think actually if we went in May, ⁓ for example, that conversation might be happening. think that's the pace of the change that we're seeing. I that was really interesting. And indeed, we're gonna get a couple of clients. We're gonna get a client from Australia and ⁓ client from ⁓ UK to talk to each other, sort ⁓ really specifically on that issue ⁓ ⁓ next couple of weeks, which is gonna be really exciting.
Tom: Yeah. Yeah. Yeah, yeah, yeah. Yeah, yeah, they'll be valuable. if there was one tangible takeaway for clients from these three parts. What would we say?
Jon: What happens next? Call it the AI revolution, call it chaos, call it whatever you like, whatever name it fundamentally gets, whatever happens next is going to be powered by data. So if you want to be part of that, and if you want to be able to demonstrate to the bill payers and the toll payers and the taxpayers and Public that, you know,
Tom: Yeah.
Jon: funds or all infrastructure eventually in some way, shape or form ⁓ want to be able to demonstrate to those people that you're doing the absolute best you can and employing best to deliver the best outcomes. You've to go back to data and really understand where you are, what quality it's got, what does it really mean and how are you going to it back into your business. for you, Tom, one takeaway from the trip.
Tom: ⁓ Take away from the trip, from every single client that we spoke to, there's a thirst to kind of take ⁓ data quality to the stage. So that's really exciting. And then I think there's an opportunity ⁓ to bring that ⁓ learning within the industries, also across industry as well.
Jon: Amazing.
Tom: Thank you everyone for joining. Thanks for tuning in. Thank you, John, for your time and your insights and for three weeks together in hotels. I don't think we had an argument, no. was amazing to talk to those clients. good to summarize that. ⁓
Jon: course. Amazing. No arguments, not a harsh word was said.
Tom: these three parts and hopefully there's a next set of episodes. Thank you very much. ⁓
Jon: Absolutely.
Tash Wilson: You've been listening to Transform by GHD. Big ideas, real impact. Welcome to Transform, tomorrow's thinking today. Real conversations on the innovation shaping businesses and our society.
Tom: heart.
Jon: I think part three to conclude, what happens next? know we've both in the UK and in Australia, a to varying degrees ⁓ data platform landscape. We've ⁓ got great stuff in some areas, some challenges. sometimes in the same areas, sometimes in other areas. got this utopia of, you know, was the AI revolution has, the phase of it ⁓ has ⁓ software, applications, platform piece becomes trivially easy to solve. got a lot of and hopefully a lot acceleration and reduced costs in that space. How do we get from where we are to actually doing that for some organizations that AI revolution is going to happen incredibly fast. Those organizations probably aren't managing safety critical infrastructure. ⁓ And we a bit about how ⁓ kind of not really possible for lot of the that our clients to move that fast.
Tom: we get there. Yeah.
Jon: But we still want to be part of that. We still need to adopt those technologies. We still want to do things that are going to take us on that journey as quickly as possible. What are those things? I think it's a really interesting question. ⁓ And a very tricky executive question in terms of how do we plan that journey in that's never been done before and is completely new for everyone. ⁓
Tom: No, no, it's not out of the book. Yeah.
Jon: There's no off the shelf plan here. do we plan out a sequence of activities and interdependent work streams that's relatively regrets what our timelines are in that. that's conversation for this morning. in terms of that planning piece, We talked a bit about making the business case for and getting that lined up. do you make the business case for a journey that is inevitable? ⁓ everyone's seen AI, everyone's, ⁓ know, you chat to you, GBT the phone. We know this is coming, but got a completely intangible value, there's no fixed timeline. what's that conversation?
Tom: Yeah. Yeah, yeah. Well, and also before the business case there's almost some managing expectations. ⁓ So not like looking at the tech industry to say, yeah, why do we not have adoption now? It a three to five journey. ⁓ But we as an need to be comfortable with that given the magnitude of the challenge, but then also the benefits of getting it right. ⁓ And then it's not kind of whizzy
Jon: Yep.
Tom: case studies and good things to talk about our conferences, although that is always good. But it's the kind of a ⁓ a build, a build, a build to ⁓ release some quite big benefits of which some of your investment across clients has already taken place. In the UK, we've seen that we worked in organisations where millions and millions was being pumped in for large systems and software where
Jon: Yeah.
Tom: benefits of those still haven't really been realized.
Jon: Yeah, absolutely. there's real kind everything's different, some but everything's somehow still the same. We're going to get that journey right ⁓ by connecting our efforts and ⁓ what we're doing to use cases deliver bottom line that deliver clear pound signs, clear dollar signs. And that sort of first principles always been true. Like we do our best work in the data and technology as a sector ⁓ when it's connected to it's not just gazing and tidying things up for the sake of it, it's connected to the ⁓ business case, the frontline value. So I think of those first principles capabilities are still going to be required to steward that journey. ⁓ But now the technology set the tool set and kind of the awareness of the horizon needs to change. And it's no longer about SQL and Python. It's about kind of understanding what the timeline is for the next large language model release and being able understand for those large language what things that we can do
Tom: Yeah, yeah, yeah, yeah,
Jon: to constrain them, to guide them, to harness them and focus them to ensure that they are delivering correct insights for our specific situation. How do teach them? How do you train them?
Tom: Yeah, yeah, yeah. And then, I mean, in some ways in this case, AI is our friend because ⁓ kind of sets ⁓ vision of where we need to go because data and systems need to be structured in a certain way for you to be able to make use of AI. And then therefore, that roadmap is set. If you don't do it in that way, then you're not going to see the benefits.
Jon: Yeah, yeah. we've spoken in both episode one and episode two, ⁓ about data quality, but the extension of that in ⁓ a AI world do understand Do we understand our own corporate logic, our tacit knowledge? Have we made our internal processes and structures machine readable in a way that we can input that to that AI workflow. I that there's a key kind of shift in the level of recursion that's now possible. Like it's no longer linear flow of source data goes in an application is transformed.
Tom: Yeah,
Jon: an insight comes out, kind of source data, AI has a go and then iterates and makes itself better 400 million times and then there's an and a very different kind of way of thinking and way of operating in ⁓ a of places.
Tom: Yeah. So what are our initial thoughts on what those steps are? And I guess with a reference to the discussions we had in Australia and the clients we're working with in the UK, some of which have similar systems, but they're regionalized as different instances of the same system. very kind pocketed data sets ⁓ is starting point, isn't it?
Jon: No time ever in planning. no time really wasted in improving your data ⁓ taking to put your data first and connect it to frontline value. So even you're not doing related to AI, efforts and projects and of. expended resource to connect your source data to frontline value, understand data quality, understand where you are and build that value chain. Even using, as I say, tooling, ⁓ still going to be worthwhile because the better that is, the faster you will be able to adopt ⁓ AI ⁓ the end of it. or when the time comes for you to be able to do that and effectively. I think the thing that we saw, we have a couple of clients here in the UK that are ⁓ expending serious on. And there were maybe a couple, maybe one in ⁓ Australia was really ready for this, but it wasn't a huge conversation that we had. How do we make our
Tom: Hmm.
Jon: corporate logic, machine readable, how do we start reaching and really understand the semantic meaning in our data and how our data sets are linked together both within the applications that we have and how do they link to reality and making that whole piece machine readable so it is ready to then constrain and guide large language models as develop insights from the data. give much ⁓ accurate insights and much more effective outputs speed. that's ⁓ kind a emerging conversation that obviously, as we've ⁓ discussed before, is an of the data quality piece. But again, this isn't something that everyone has already done or lots of people have already done. Some people have done it. There's some really interesting work in the space. there's no Wikipedia article for this Right. So there's a lot of exploration to do. But it's exploration that's going to have value because it's ⁓ not huge cost in the grand scheme of It's thinking big thoughts and really, ⁓ you know, getting to grips with what
Tom: Yeah. Yeah, yeah.
Jon: do we really do and as a business and as a set of people and as an organization and how does our data to that. But that's going to have immense value when the time comes for us to start in the AI tools.
Tom: Yeah, yeah, yeah. And then I don't know if we skirted over this a little bit, but to the ⁓ the business case. we've examples in the where organizations have tried to kind of value the data if it's at certain quality that the business case hangs off that. Is that the way to do it? other approaches.
Jon: think that particular case, and it was interesting because that was a client in the UK that had ⁓ a really extensive data valuation. And was a number of clients in Australia that were aware of that and ⁓ had to that or understand how they could do something similar. think what's probably important, the pace change that we're currently seeing, is that
Tom: Yeah.
Jon: the business case simply made out of fear of if we don't try and sprint at this, we're going to get somehow left behind or get heavily disrupted off the of the planet. That's not the right answer. The right answer is.
Tom: you
Jon: ⁓ getting a good understanding of what the future is going to look like, getting a coherent plan and going on the journey because with AI, similarly all the applications and platforms and all the bits of technology that you could buy previous, there's still going to be the pitfall. There's still going to be the opportunity to expend a bunch of effort, do a bunch of stuff not really make any sustainable progress or have a misfire with the business because we've done something but we didn't bring the people along with us. all ⁓ those same pitfalls are still going to be there. So it's to be possible to try too hard.
Tom: Yeah, yeah. ⁓ ⁓ an opportunity here that it's, you don't need to procure a massive multi-year contract with a big single supplier. Meaning that you could have put your eggs and your cost all in one basket that actually you can have that roadmap and that plan, but kind of stage gated. So you're, you know, we're doing this, this is effort that we need to put in. This is the support we need from other organizations.
Jon: Hmm.
Tom: then we'll get to here and then we'll, we think we know the next step but we'll take stock and then we'll recommission from that point.
Jon: that's an area where the more mature clients that we have here the UK that really are focusing on surfacing corporate meaning and putting their data first and their data at the heart of rather than their applications, going to be able to the best use of that. ⁓ data is going to drive and then the application piece becomes a triviality and vendor lock-in could potentially become a thing the past. We didn't talk to anyone that had got to that stage yet ⁓ in Australia. And I think actually if we went in May, ⁓ for example, that conversation might be happening. think that's the pace of the change that we're seeing. I that was really interesting. And indeed, we're gonna get a couple of clients. We're gonna get a client from Australia and ⁓ client from ⁓ UK to talk to each other, sort ⁓ really specifically on that issue ⁓ ⁓ next couple of weeks, which is gonna be really exciting.
Tom: Yeah. Yeah. Yeah, yeah, yeah. Yeah, yeah, they'll be valuable. if there was one tangible takeaway for clients from these three parts. What would we say?
Jon: What happens next? Call it the AI revolution, call it chaos, call it whatever you like, whatever name it fundamentally gets, whatever happens next is going to be powered by data. So if you want to be part of that, and if you want to be able to demonstrate to the bill payers and the toll payers and the taxpayers and Public that, you know,
Tom: Yeah.
Jon: funds or all infrastructure eventually in some way, shape or form ⁓ want to be able to demonstrate to those people that you're doing the absolute best you can and employing best to deliver the best outcomes. You've to go back to data and really understand where you are, what quality it's got, what does it really mean and how are you going to it back into your business. for you, Tom, one takeaway from the trip.
Tom: ⁓ Take away from the trip, from every single client that we spoke to, there's a thirst to kind of take ⁓ data quality to the stage. So that's really exciting. And then I think there's an opportunity ⁓ to bring that ⁓ learning within the industries, also across industry as well.
Jon: Amazing.
Tom: Thank you everyone for joining. Thanks for tuning in. Thank you, John, for your time and your insights and for three weeks together in hotels. I don't think we had an argument, no. was amazing to talk to those clients. good to summarize that. ⁓
Jon: course. Amazing. No arguments, not a harsh word was said.
Tom: these three parts and hopefully there's a next set of episodes. Thank you very much. ⁓
Jon: Absolutely.
