1 00:00:08,760 --> 00:00:11,799 Speaker 1: Welcome to the Business of Tech powered by two Degrees Business. 2 00:00:11,840 --> 00:00:15,160 Speaker 1: I'm Peter Griffin and in this episode I'm talking to 3 00:00:15,280 --> 00:00:19,520 Speaker 1: Greg Davidson, the man responsible for leading the country's biggest 4 00:00:19,560 --> 00:00:24,959 Speaker 1: IT company, which employs five thousand workers across Australasia and 5 00:00:25,040 --> 00:00:29,240 Speaker 1: through its payroll platforms, is responsible for paying around half 6 00:00:29,280 --> 00:00:35,559 Speaker 1: a million Kiwis every fortnight. Datacom is celebrating sixty years 7 00:00:35,560 --> 00:00:36,320 Speaker 1: in business. 8 00:00:36,760 --> 00:00:38,000 Speaker 2: It's a true. 9 00:00:37,760 --> 00:00:42,000 Speaker 1: Kiwi success story, powering some of the biggest digital transformations 10 00:00:42,000 --> 00:00:44,639 Speaker 1: in our region in a private sector as well as 11 00:00:44,640 --> 00:00:48,159 Speaker 1: in government. It has a roster of big clients across 12 00:00:48,240 --> 00:00:52,159 Speaker 1: the Tasman as well. Datacom was founded back in nineteen 13 00:00:52,280 --> 00:00:53,840 Speaker 1: sixty five in christ Church. 14 00:00:53,920 --> 00:00:55,640 Speaker 2: It's weathered every. 15 00:00:55,400 --> 00:00:59,920 Speaker 1: Major technological wave since the rise of computers, from mainframes 16 00:01:00,040 --> 00:01:04,240 Speaker 1: to the Internet, mobile devices, the cloud of course, and 17 00:01:04,280 --> 00:01:09,120 Speaker 1: now artificial intelligence. As Greg Davidson, who is Datacom's group 18 00:01:09,280 --> 00:01:13,000 Speaker 1: CEO and has led the company for eighteen years, shares 19 00:01:13,200 --> 00:01:18,200 Speaker 1: in this interview, Datacom faces another technological turning point with 20 00:01:18,360 --> 00:01:23,240 Speaker 1: AI navigating the risks and extraordinary opportunities of the technology 21 00:01:23,280 --> 00:01:27,080 Speaker 1: at a time when every organization large and small faces 22 00:01:27,240 --> 00:01:32,559 Speaker 1: unprecedented change It's a wide ranging conversation about tech, about 23 00:01:32,640 --> 00:01:36,800 Speaker 1: leadership and what comes next for both Datacom and New Zealand. 24 00:01:37,400 --> 00:01:40,320 Speaker 1: Here's the interview with Datacom's Greg Davidson. 25 00:01:46,959 --> 00:01:49,680 Speaker 2: Greg Davidson, Welcome to the Business of Tech. How are 26 00:01:49,680 --> 00:01:50,040 Speaker 2: you doing? 27 00:01:50,560 --> 00:01:53,080 Speaker 3: Very good, Thanks, Peter. I'm really pleased to be here. 28 00:01:53,200 --> 00:01:58,160 Speaker 1: Yeah, and it's a big year for Datacom. Datacom turns 29 00:01:58,400 --> 00:02:03,280 Speaker 1: sixty this year. Founded in nineteen sixty five, our biggest 30 00:02:03,440 --> 00:02:07,520 Speaker 1: New Zealand owned IT company spans Australia and New Zealand. 31 00:02:07,560 --> 00:02:07,920 Speaker 3: One of the. 32 00:02:07,840 --> 00:02:14,480 Speaker 2: Biggest employers in Australasia. Incredible company, incredible success. You've been there, 33 00:02:14,520 --> 00:02:17,720 Speaker 2: This is pretty amazing. Nearly half of that time, twenty 34 00:02:17,760 --> 00:02:22,079 Speaker 2: eight years. You joined Datacom in nineteen ninety seven, became 35 00:02:22,200 --> 00:02:25,760 Speaker 2: the chief executive officer in two thousand and seven, which 36 00:02:25,840 --> 00:02:28,520 Speaker 2: is crazy to think about it the year that iPhone 37 00:02:28,760 --> 00:02:34,720 Speaker 2: was released. So this company has seen numerous waves of 38 00:02:35,000 --> 00:02:37,840 Speaker 2: technological change. You've been there basically for the rise of 39 00:02:37,880 --> 00:02:42,919 Speaker 2: the Internet, the rise of smartphones and apps, the cloud revolution, 40 00:02:43,080 --> 00:02:46,080 Speaker 2: and now that AI revolution. We're going to talk about 41 00:02:46,120 --> 00:02:48,080 Speaker 2: all of that, but maybe just take us back to 42 00:02:48,160 --> 00:02:51,880 Speaker 2: your personal journey into datacom what you grew up in 43 00:02:51,919 --> 00:02:54,560 Speaker 2: the Hut Valley and then went into what computer science 44 00:02:54,560 --> 00:02:55,200 Speaker 2: at Victoria. 45 00:02:55,360 --> 00:02:58,280 Speaker 3: Yeah, that's right, if you want the slightly longer version 46 00:02:58,280 --> 00:03:00,800 Speaker 3: of it. I had two parents who are teacher, so 47 00:03:00,880 --> 00:03:03,800 Speaker 3: I grew up in Hawk's Bay in Parmerston North and 48 00:03:03,840 --> 00:03:05,600 Speaker 3: then there was a journey down to Wellington at one 49 00:03:05,600 --> 00:03:08,000 Speaker 3: point where she moved between the two, and then it 50 00:03:08,160 --> 00:03:11,639 Speaker 3: was high school in the Heart Valley. Yeah. Data Comm 51 00:03:11,680 --> 00:03:14,280 Speaker 3: turning sixty, it is older than me. I just I'm 52 00:03:14,520 --> 00:03:16,000 Speaker 3: going to hang on to that for as long as 53 00:03:17,520 --> 00:03:20,960 Speaker 3: it began in the bureau era and has been through many, many, 54 00:03:21,000 --> 00:03:23,720 Speaker 3: many changes, and I think as we sort of expand 55 00:03:23,720 --> 00:03:26,000 Speaker 3: on the conversations we've got today about the different areas 56 00:03:26,000 --> 00:03:28,560 Speaker 3: of computing, you'll find that data Comm's had to adapt 57 00:03:28,600 --> 00:03:31,200 Speaker 3: through it. My personal journey, I'm a lapsed software engineer, 58 00:03:31,240 --> 00:03:34,800 Speaker 3: so I studied software engineering at Victoria in Canterbury and 59 00:03:34,880 --> 00:03:40,800 Speaker 3: then went through a kind of small business doing lots 60 00:03:40,800 --> 00:03:44,160 Speaker 3: of small jobs for small business owners into working at 61 00:03:44,160 --> 00:03:48,360 Speaker 3: the point where computers really impacted design, and then multimedia 62 00:03:48,400 --> 00:03:50,440 Speaker 3: and then the Internet, and then my journey took me 63 00:03:50,520 --> 00:03:54,160 Speaker 3: to dot com A couple of years after the Internet 64 00:03:54,280 --> 00:03:56,280 Speaker 3: really exploded across the world. 65 00:03:56,480 --> 00:03:58,240 Speaker 1: Yeah, and that was a huge area of growth for 66 00:03:58,360 --> 00:04:00,240 Speaker 1: data com You mentioned that, you know, it came out 67 00:04:00,240 --> 00:04:03,080 Speaker 1: of the bureau era, so that was nineteen sixty five. 68 00:04:03,160 --> 00:04:07,680 Speaker 1: The Computer Bureau Limited Company was the precursor company to 69 00:04:07,960 --> 00:04:12,440 Speaker 1: data com. This was the founders Paul Hargreaves, doctor Bernard 70 00:04:12,480 --> 00:04:16,320 Speaker 1: Battersby I think was an economics lecturer at Cannabi University 71 00:04:16,360 --> 00:04:19,719 Speaker 1: at the time. Paul was an accountant. They decided to 72 00:04:19,960 --> 00:04:23,200 Speaker 1: let's buy one of these early sort of really powerful computers. 73 00:04:23,200 --> 00:04:25,920 Speaker 1: It was the ICL nineteen oh two, it was called. 74 00:04:25,960 --> 00:04:29,640 Speaker 1: They raised thirty thousand pounds at the time from shareholders. 75 00:04:29,800 --> 00:04:31,880 Speaker 1: What did they start out doing with this sort of 76 00:04:31,960 --> 00:04:33,920 Speaker 1: bureau approach to computing? 77 00:04:34,000 --> 00:04:36,680 Speaker 3: The answer to your question is a superlogical one. What's 78 00:04:36,800 --> 00:04:40,080 Speaker 3: the thing that requires lots of calculations that you would 79 00:04:40,080 --> 00:04:43,880 Speaker 3: buy a computer and share it across multiple companies to 80 00:04:43,920 --> 00:04:46,680 Speaker 3: do in that era? Payroll? Yeah, you know, we're still 81 00:04:46,680 --> 00:04:49,640 Speaker 3: a payroll provider now. We have a very modern fac 82 00:04:49,680 --> 00:04:51,719 Speaker 3: two very modern versions of that and Smartly and data 83 00:04:51,760 --> 00:04:55,159 Speaker 3: pay and we pay about half a million key weeks 84 00:04:55,200 --> 00:04:59,279 Speaker 3: every fortnite using those platforms, and in that would be 85 00:04:59,320 --> 00:05:03,000 Speaker 3: over thirty thousand small to medium businesses. So if you 86 00:05:03,000 --> 00:05:05,640 Speaker 3: want that original starting point back in the sixties, it's 87 00:05:05,880 --> 00:05:09,440 Speaker 3: iterated several times. We've invested heavily in a modern generation 88 00:05:09,520 --> 00:05:11,640 Speaker 3: of it and it still exists as a small piece 89 00:05:11,640 --> 00:05:14,320 Speaker 3: of what we do. Not everybody knows because of course 90 00:05:14,320 --> 00:05:16,760 Speaker 3: Smartly operates under its own brand and deals with a 91 00:05:16,760 --> 00:05:18,520 Speaker 3: lot of small business, which is a bit different to 92 00:05:18,600 --> 00:05:21,200 Speaker 3: what the data com brand is usually associated with. The 93 00:05:21,240 --> 00:05:21,960 Speaker 3: New Zealand. 94 00:05:21,760 --> 00:05:26,080 Speaker 1: Payroll has been the constant throughout and on both sides 95 00:05:26,120 --> 00:05:26,640 Speaker 1: of the Tasman. 96 00:05:26,720 --> 00:05:28,240 Speaker 2: Now you do payroll. 97 00:05:28,839 --> 00:05:31,600 Speaker 1: In the seventies it was the introduction of Cobol for 98 00:05:32,040 --> 00:05:35,800 Speaker 1: payroll systems. You were integral to that, bringing Oracle to 99 00:05:35,839 --> 00:05:39,920 Speaker 1: New zealand getting big into databases as well. So there's 100 00:05:39,960 --> 00:05:43,279 Speaker 1: been quite a few milestones there. What was it like 101 00:05:43,520 --> 00:05:46,760 Speaker 1: your conversations with the likes of Paul Hargreaves, John Holsworth, 102 00:05:46,760 --> 00:05:50,080 Speaker 1: who was a shareholder then became the chairman of Data 103 00:05:50,080 --> 00:05:53,200 Speaker 1: Common in the late eighties. You know, last year I 104 00:05:53,240 --> 00:05:58,560 Speaker 1: went to Hewlett Packard in Palo Alto and they've preserved 105 00:05:58,560 --> 00:06:02,240 Speaker 1: the office there where Bill Hewlett and Dave Packard used 106 00:06:02,240 --> 00:06:04,760 Speaker 1: to sit and they positioned their desks so they could 107 00:06:04,920 --> 00:06:06,640 Speaker 1: had a line of site with each other. So if 108 00:06:06,680 --> 00:06:08,080 Speaker 1: one of them was on the phone trying to do 109 00:06:08,120 --> 00:06:10,520 Speaker 1: a deal, he could gestured to the other guy to 110 00:06:10,560 --> 00:06:13,360 Speaker 1: come over and join the conversation if we needed him 111 00:06:13,400 --> 00:06:15,600 Speaker 1: to get the deal over the line. And I think 112 00:06:15,640 --> 00:06:18,880 Speaker 1: in New Zealand, the likes of Paul High Greaves, Bernard 113 00:06:18,920 --> 00:06:22,480 Speaker 1: battersby John Holdsworth sort of have that same sort of legacy. 114 00:06:22,520 --> 00:06:25,960 Speaker 1: They were some of the pioneers of our tech sector. 115 00:06:26,040 --> 00:06:28,440 Speaker 1: But interested in your interactions with them and what you've 116 00:06:28,560 --> 00:06:30,400 Speaker 1: learned from them over the years. 117 00:06:30,480 --> 00:06:33,880 Speaker 3: I met Paul obviously and had plenty of opportunity to 118 00:06:33,920 --> 00:06:36,960 Speaker 3: interact with them. So I joined during the era when 119 00:06:37,279 --> 00:06:41,080 Speaker 3: John was checked in the late nineties and Frank Stevenson 120 00:06:41,480 --> 00:06:43,839 Speaker 3: was CEO. You know, if you look at other founders 121 00:06:43,920 --> 00:06:46,480 Speaker 3: that exist around the world, or people who really kind 122 00:06:46,480 --> 00:06:49,000 Speaker 3: of put their DNA into a company, John's was worked 123 00:06:49,000 --> 00:06:53,240 Speaker 3: back from customer. It was a essential to a services organization, 124 00:06:53,360 --> 00:06:55,440 Speaker 3: the reputation for what you can do in the market 125 00:06:55,600 --> 00:06:58,160 Speaker 3: and the ability to find the intersection of what your 126 00:06:58,160 --> 00:07:01,359 Speaker 3: customers need and what we're good at or what we 127 00:07:01,440 --> 00:07:03,760 Speaker 3: must become in order to be able to fulfill it 128 00:07:03,839 --> 00:07:06,520 Speaker 3: I still think is alive in what we are today, 129 00:07:06,600 --> 00:07:09,080 Speaker 3: and i'd very much pointed out there was another piece 130 00:07:09,160 --> 00:07:11,360 Speaker 3: that was to do with you know, some of what 131 00:07:11,400 --> 00:07:13,720 Speaker 3: you do for customers is very specific to them. But 132 00:07:13,760 --> 00:07:16,520 Speaker 3: on the other hand, there's some things where New Zealand's 133 00:07:16,560 --> 00:07:19,640 Speaker 3: relatively small and finding economy of scale can be quite 134 00:07:19,640 --> 00:07:22,080 Speaker 3: a challenging thing. And so a lot of what we 135 00:07:22,160 --> 00:07:24,559 Speaker 3: do in the New Zealand market and the Australian market 136 00:07:24,600 --> 00:07:26,600 Speaker 3: is where customers can't find their own economy of scale 137 00:07:26,600 --> 00:07:27,960 Speaker 3: and doing it, and therefore we can do it on 138 00:07:27,960 --> 00:07:30,520 Speaker 3: a shared basis for them that makes sense. And then 139 00:07:30,560 --> 00:07:32,800 Speaker 3: there's another piece of it, of course, which is the 140 00:07:32,880 --> 00:07:38,400 Speaker 3: coding software development, build systems for either broadly to meet 141 00:07:38,480 --> 00:07:41,280 Speaker 3: need in the market or very specifically for customers businesses. 142 00:07:41,360 --> 00:07:44,880 Speaker 1: And you've definitely achieved that economies of scale. Just recently 143 00:07:44,920 --> 00:07:49,280 Speaker 1: you put out your financial results for the full year 144 00:07:50,480 --> 00:07:54,040 Speaker 1: and one point four eight billion in revenue, and you 145 00:07:54,080 --> 00:07:56,000 Speaker 1: know you've been over the billion dollar MARC for several 146 00:07:56,080 --> 00:07:59,240 Speaker 1: years now, so this is a huge scale you've got there, 147 00:07:59,520 --> 00:08:01,560 Speaker 1: thirty seven million dollar profit. I guess it was a 148 00:08:01,560 --> 00:08:06,240 Speaker 1: pretty tough year last year, so good to be profitable 149 00:08:06,280 --> 00:08:09,920 Speaker 1: and to be reducing your debt and growing revenue as well. 150 00:08:10,000 --> 00:08:12,680 Speaker 3: Because we operate in New Zealand and Australia, and because 151 00:08:12,720 --> 00:08:16,400 Speaker 3: we operate across quite a lot of different kinds of activities, 152 00:08:16,560 --> 00:08:19,720 Speaker 3: the balance between the two does give us some ability 153 00:08:19,720 --> 00:08:22,520 Speaker 3: to cope a wee bit better with market fluctuations and 154 00:08:22,560 --> 00:08:25,320 Speaker 3: certainly the fact that New Zealand's economy is really hurting 155 00:08:25,360 --> 00:08:26,960 Speaker 3: at the moment and it's taking a long time for 156 00:08:26,960 --> 00:08:30,880 Speaker 3: business confidence to come back, and so every organization quite 157 00:08:30,880 --> 00:08:33,520 Speaker 3: sensibly during that period it's focused on value for money, 158 00:08:34,040 --> 00:08:38,240 Speaker 3: wants what's done to have a path to benefit very quickly. 159 00:08:38,800 --> 00:08:40,360 Speaker 3: And of course at the opposite end of the spectrum, 160 00:08:40,400 --> 00:08:43,720 Speaker 3: that's very much challenged by the fact that the AI revolution, 161 00:08:43,880 --> 00:08:45,960 Speaker 3: i'll call that deliberately and perhaps and maybe even a 162 00:08:45,960 --> 00:08:49,959 Speaker 3: bit provocatively, offers just such a huge change in what's 163 00:08:50,000 --> 00:08:54,280 Speaker 3: possible with technology, but it does require that you very 164 00:08:54,320 --> 00:08:56,880 Speaker 3: actively pursue those benefits. It's not just going to happen 165 00:08:56,960 --> 00:08:59,960 Speaker 3: because you buy a product. It's going to happen because 166 00:09:00,360 --> 00:09:02,840 Speaker 3: you really focus on where you can find the benefits 167 00:09:02,840 --> 00:09:05,599 Speaker 3: and make good choices as to where within new organization 168 00:09:05,679 --> 00:09:08,240 Speaker 3: and can make a difference. So and you know, we've 169 00:09:08,240 --> 00:09:11,520 Speaker 3: got our own version of that. We've got multiple examples 170 00:09:11,559 --> 00:09:14,560 Speaker 3: of how we're adapting what we do internally and going 171 00:09:14,600 --> 00:09:18,520 Speaker 3: after automation benefits and artificial intelligence field benefits, and we've 172 00:09:18,520 --> 00:09:19,920 Speaker 3: got a whole lot of activity. We've got to the 173 00:09:20,000 --> 00:09:22,440 Speaker 3: market with our customers too, where we're helping them adopt. 174 00:09:22,679 --> 00:09:25,320 Speaker 1: You've recently sort of restructured the business, taken a bit 175 00:09:25,360 --> 00:09:29,560 Speaker 1: of a trim on headcount, which most large tech companies have, 176 00:09:29,720 --> 00:09:31,559 Speaker 1: So talk about that in a minute, but just before 177 00:09:31,559 --> 00:09:35,960 Speaker 1: we do the key to that stability and the ability 178 00:09:36,000 --> 00:09:38,120 Speaker 1: to stay a large company where there are a lot 179 00:09:38,160 --> 00:09:41,640 Speaker 1: of competitors. You've got the multinational consultancy firms are very 180 00:09:41,640 --> 00:09:43,520 Speaker 1: active in New Zealand. You've got the likes of Spark 181 00:09:44,080 --> 00:09:46,800 Speaker 1: and then a lot of IT integrators in the market 182 00:09:46,840 --> 00:09:49,160 Speaker 1: as well. But you seem to be able to over 183 00:09:49,240 --> 00:09:54,720 Speaker 1: decades maintain this consistent revenues and loyal customers over the 184 00:09:54,920 --> 00:09:57,679 Speaker 1: decades as well. How do you do that but also 185 00:09:58,000 --> 00:10:02,719 Speaker 1: be innovative and moving and adapt to emerging technologies. 186 00:10:02,880 --> 00:10:05,920 Speaker 3: If you look at market share data, particularly of the 187 00:10:05,960 --> 00:10:08,560 Speaker 3: systems in the greater piece of what we are systems 188 00:10:08,559 --> 00:10:11,080 Speaker 3: into greater outsource, so you know the services part, which 189 00:10:11,120 --> 00:10:13,520 Speaker 3: is what Datacon and the data com brand's best known for. 190 00:10:13,600 --> 00:10:15,920 Speaker 3: In both New Zealand and Australia, you're going to find 191 00:10:16,480 --> 00:10:20,480 Speaker 3: that in the landscape of top ten organizations by size. 192 00:10:20,600 --> 00:10:23,120 Speaker 3: But that's both revenue and headcount in Australia and New 193 00:10:23,160 --> 00:10:26,720 Speaker 3: Zealand combined. We're in the top ten. We're pushing our 194 00:10:26,760 --> 00:10:28,679 Speaker 3: way slowly up it, and I've been doing that over 195 00:10:28,720 --> 00:10:30,920 Speaker 3: a long period of time that we're in the top five. 196 00:10:31,280 --> 00:10:35,560 Speaker 3: Not that that scale and market share is necessary an 197 00:10:35,600 --> 00:10:39,200 Speaker 3: answering and of itself, but every competitor is either a 198 00:10:39,200 --> 00:10:41,480 Speaker 3: global organization or a tel co in that list. So 199 00:10:41,480 --> 00:10:43,640 Speaker 3: we're the only pure play owned in this part of 200 00:10:43,679 --> 00:10:48,400 Speaker 3: the world services organization that organizations can look to for 201 00:10:49,120 --> 00:10:51,760 Speaker 3: impartial advice about the technology waves that they are hitting, 202 00:10:51,840 --> 00:10:55,800 Speaker 3: for what you should use for what, for assistance in 203 00:10:55,880 --> 00:10:59,400 Speaker 3: achieving their goals. And I think that position we consider 204 00:10:59,440 --> 00:11:02,400 Speaker 3: it to be. I describe it as a responsibility in 205 00:11:02,440 --> 00:11:05,000 Speaker 3: that there's so much selling that goes on in the 206 00:11:05,000 --> 00:11:08,240 Speaker 3: tech space that doesn't necessarily have what the customers need 207 00:11:08,760 --> 00:11:13,160 Speaker 3: at its core. That in order for us to maintain 208 00:11:13,240 --> 00:11:17,480 Speaker 3: that over time, we have to be able to really 209 00:11:17,559 --> 00:11:19,880 Speaker 3: hear what it is that's the customer's priorities, then get 210 00:11:20,000 --> 00:11:24,240 Speaker 3: really good at matching the solutions we build for them 211 00:11:24,360 --> 00:11:27,679 Speaker 3: or offer them to that need. It's difficult, and particularly 212 00:11:27,679 --> 00:11:30,360 Speaker 3: in times of inflection as you pointed to some of 213 00:11:30,360 --> 00:11:32,440 Speaker 3: the big ones at the beginning when you opened up. 214 00:11:32,440 --> 00:11:35,640 Speaker 3: You know, you talked about the Internet era, the mobile era, 215 00:11:36,240 --> 00:11:38,439 Speaker 3: the cloud era, and now the AI era, and they 216 00:11:38,480 --> 00:11:41,680 Speaker 3: are four big periods of change in the industry. Now, 217 00:11:42,120 --> 00:11:44,480 Speaker 3: how do we do it? Look the difficult bit And 218 00:11:44,520 --> 00:11:46,360 Speaker 3: I'm sure if you talk to any of the team 219 00:11:46,400 --> 00:11:48,640 Speaker 3: and data comm they'd say they see this in me. 220 00:11:49,200 --> 00:11:52,360 Speaker 3: You can't ever stand still, and you've got to be 221 00:11:52,400 --> 00:11:55,240 Speaker 3: willing to relentlessly pursue what you need to be, not 222 00:11:55,280 --> 00:11:58,880 Speaker 3: what you are now. And that's really uncomfortable. Being willing 223 00:11:59,000 --> 00:12:03,760 Speaker 3: to change the place quickly when technology revolutions land, being 224 00:12:03,800 --> 00:12:07,600 Speaker 3: willing to really try and work out out of all 225 00:12:07,600 --> 00:12:10,959 Speaker 3: of the selling that's been fired by global platforms. Being 226 00:12:11,000 --> 00:12:13,520 Speaker 3: able to choose what's best for a customer in any 227 00:12:13,520 --> 00:12:16,560 Speaker 3: given circumstances. How to do it economically is actually a 228 00:12:16,600 --> 00:12:18,439 Speaker 3: really complicated skill because you've got to get past the 229 00:12:18,480 --> 00:12:20,360 Speaker 3: marketing and the hyperse You've actually got to figure out 230 00:12:20,400 --> 00:12:24,480 Speaker 3: how can I reliably and repetitively do this across our 231 00:12:24,520 --> 00:12:25,120 Speaker 3: customer base. 232 00:12:25,200 --> 00:12:27,480 Speaker 1: Will you put your finger on what I think is 233 00:12:27,520 --> 00:12:32,319 Speaker 1: one of the real reasons for data Comm's successes your independence. 234 00:12:32,440 --> 00:12:34,600 Speaker 1: I mean, there are tech companies in New Zealand that 235 00:12:35,440 --> 00:12:39,320 Speaker 1: they are Microsoft integrators. They live and breathe Microsoft. You know, 236 00:12:39,320 --> 00:12:42,080 Speaker 1: they're a partner for you, but you partner with Google Aws, 237 00:12:42,640 --> 00:12:45,600 Speaker 1: you know everyone. So when a government department, you know, 238 00:12:45,640 --> 00:12:49,040 Speaker 1: looking for a high trust tech partner, goes to datacom, 239 00:12:49,240 --> 00:12:51,439 Speaker 1: you're not going to be pushing one particular vendor. You're 240 00:12:51,440 --> 00:12:53,680 Speaker 1: going to be looking at what the customer needs and 241 00:12:53,720 --> 00:12:56,840 Speaker 1: then finding the best technology to suit it. So I 242 00:12:56,840 --> 00:13:01,000 Speaker 1: think it's that tech agnostic sort of approach that underpend 243 00:13:01,400 --> 00:13:01,880 Speaker 1: your growth. 244 00:13:01,920 --> 00:13:04,000 Speaker 3: I think it's a really good summary of it, and 245 00:13:04,080 --> 00:13:07,240 Speaker 3: actually one of the really interesting things. And you know, 246 00:13:07,240 --> 00:13:10,839 Speaker 3: if you've been following the AI revolution closely, you'll see 247 00:13:10,880 --> 00:13:13,760 Speaker 3: that the frontier companies in the AI space aren't by 248 00:13:13,800 --> 00:13:17,480 Speaker 3: and large your traditional players, and so you've actually got 249 00:13:17,520 --> 00:13:22,559 Speaker 3: a really interesting situation where new organizations are rising all 250 00:13:22,600 --> 00:13:25,280 Speaker 3: of the big established players in the market with out 251 00:13:25,280 --> 00:13:27,760 Speaker 3: of brand names that you know in technology that every 252 00:13:27,880 --> 00:13:31,240 Speaker 3: house just about every household knows, are all suddenly in 253 00:13:31,280 --> 00:13:34,240 Speaker 3: an immediate furious set of investment in R and D 254 00:13:34,360 --> 00:13:36,600 Speaker 3: in activity in order to figure out how to catch 255 00:13:36,679 --> 00:13:39,040 Speaker 3: up with this revolution that's just sort of turned up 256 00:13:39,080 --> 00:13:41,840 Speaker 3: over the last couple of years. You've got frontier companies 257 00:13:42,040 --> 00:13:46,720 Speaker 3: like open Ai and Anthropic, and i'd put probably Google's 258 00:13:46,720 --> 00:13:49,440 Speaker 3: Gemini in that max as well. You've got the rise 259 00:13:49,480 --> 00:13:53,760 Speaker 3: of video and a MD as part of that conversation too. 260 00:13:54,280 --> 00:13:58,200 Speaker 3: This is something that every commercial organization does need to 261 00:13:58,320 --> 00:14:02,920 Speaker 3: understand because when you're in one of these disruptive periods 262 00:14:02,960 --> 00:14:05,439 Speaker 3: where your traditional players may or may not have the answer, 263 00:14:05,679 --> 00:14:08,080 Speaker 3: you've then got to go searching for the best solution 264 00:14:08,360 --> 00:14:10,640 Speaker 3: for the problem. And you've got to be really careful 265 00:14:10,760 --> 00:14:15,760 Speaker 3: to mix engineering understanding and commercial understanding and cost effectiveness 266 00:14:15,880 --> 00:14:18,320 Speaker 3: understanding to ensure that you get the upcomes you need. 267 00:14:18,679 --> 00:14:21,640 Speaker 3: And so we think in times like this, firstly, we 268 00:14:21,680 --> 00:14:24,240 Speaker 3: need to learn like crazy and really inform ourselves. Secondly, 269 00:14:24,280 --> 00:14:26,680 Speaker 3: we need to sort of be customer zero of the change. 270 00:14:27,200 --> 00:14:29,080 Speaker 3: And third we then need to figure out how to 271 00:14:29,120 --> 00:14:30,920 Speaker 3: distill some learnings of that. And there's a ton we 272 00:14:30,920 --> 00:14:33,320 Speaker 3: could talk about in this in that space out to 273 00:14:33,440 --> 00:14:35,760 Speaker 3: our customer base to help them make wise decisions. On 274 00:14:35,760 --> 00:14:36,600 Speaker 3: the other side of it. 275 00:14:36,640 --> 00:14:39,880 Speaker 1: You and your team, you senior leadership team, have obviously 276 00:14:39,920 --> 00:14:43,080 Speaker 1: been looking at the growth of AI and that the 277 00:14:43,240 --> 00:14:47,040 Speaker 1: error of generative AI large language models in particular, and 278 00:14:47,160 --> 00:14:51,000 Speaker 1: so if we're to best grab this opportunity, we're going 279 00:14:51,080 --> 00:14:53,320 Speaker 1: to have to change how we do things. So talk 280 00:14:53,400 --> 00:14:55,600 Speaker 1: us through this sort of restructure you've done recently and 281 00:14:55,640 --> 00:14:58,760 Speaker 1: how that's positioning you for the era of AI. 282 00:14:58,920 --> 00:15:03,040 Speaker 3: The senior leadership team is comprised of the heads of 283 00:15:03,560 --> 00:15:06,720 Speaker 3: our two primary markets, which are New Zealand and Australia, 284 00:15:07,000 --> 00:15:09,320 Speaker 3: and they're the folk who are in the face of 285 00:15:09,360 --> 00:15:12,880 Speaker 3: our customers, ensuring that we choose the right things that 286 00:15:12,920 --> 00:15:15,400 Speaker 3: we do to put in front of them, ensuring that 287 00:15:15,440 --> 00:15:17,920 Speaker 3: the work we do in that country is meaningful for 288 00:15:18,040 --> 00:15:19,800 Speaker 3: the country, which is a core thing that we want 289 00:15:19,840 --> 00:15:23,080 Speaker 3: to achieve, and then our capabilities that in essence, the 290 00:15:23,120 --> 00:15:27,520 Speaker 3: things we're good at. We made sure operated on across 291 00:15:27,600 --> 00:15:30,320 Speaker 3: everything data Com does basis rather than could be one 292 00:15:30,320 --> 00:15:33,240 Speaker 3: country or the other country, because we felt that the 293 00:15:33,320 --> 00:15:35,400 Speaker 3: customer demand was put the best of what you can 294 00:15:35,440 --> 00:15:38,040 Speaker 3: do in front of us every time, regardless of which 295 00:15:38,080 --> 00:15:40,760 Speaker 3: city had happened to be in, and so we needed 296 00:15:40,800 --> 00:15:43,600 Speaker 3: to organize differently in order to achieve that. And then 297 00:15:43,960 --> 00:15:47,080 Speaker 3: where it is helping us is we obviously need to 298 00:15:47,120 --> 00:15:49,520 Speaker 3: iterate and change what we do in the market to 299 00:15:49,560 --> 00:15:52,160 Speaker 3: reflect all of the capabilities that general who of AI 300 00:15:52,240 --> 00:15:54,680 Speaker 3: have added into the kitbag. We're able to do that 301 00:15:54,720 --> 00:15:57,640 Speaker 3: far more productively and effectively having made the shift in 302 00:15:57,720 --> 00:16:00,800 Speaker 3: how we operate. There's no doubt that by the time 303 00:16:00,840 --> 00:16:03,560 Speaker 3: you add it to the kind of machine learning capability 304 00:16:03,600 --> 00:16:05,880 Speaker 3: and image recognition and speech recognition and all the other 305 00:16:05,920 --> 00:16:09,960 Speaker 3: pieces of the AI equation offers the opportunity for anything 306 00:16:09,960 --> 00:16:14,440 Speaker 3: that involves document handling or words to be done very differently. 307 00:16:14,640 --> 00:16:17,400 Speaker 3: And while some perhaps of the uses in the early 308 00:16:17,440 --> 00:16:19,520 Speaker 3: era of it, which are you know, I'm going to 309 00:16:19,600 --> 00:16:24,280 Speaker 3: chat with a GENAI system obvious, the latter stages of 310 00:16:24,280 --> 00:16:26,280 Speaker 3: it are happening, the next couple of stages of it 311 00:16:26,280 --> 00:16:29,360 Speaker 3: are happening really quickly and hot off the heels of 312 00:16:29,360 --> 00:16:31,960 Speaker 3: this the building of expert systems, And one of our 313 00:16:32,000 --> 00:16:34,680 Speaker 3: examples is we've built a going back to that payroll 314 00:16:34,760 --> 00:16:36,520 Speaker 3: use case. It's not all we do, but it's worth, 315 00:16:37,440 --> 00:16:39,040 Speaker 3: it's really worth coming back to this one. So the 316 00:16:39,080 --> 00:16:41,760 Speaker 3: team there have built payroll advisor that can pass the 317 00:16:41,760 --> 00:16:44,960 Speaker 3: Payroll Practitioners exam. To build that, what you need to 318 00:16:45,000 --> 00:16:47,520 Speaker 3: do is you need to do a couple of things 319 00:16:47,560 --> 00:16:49,520 Speaker 3: on top of what you'd get out of one of 320 00:16:49,560 --> 00:16:51,960 Speaker 3: the large language models. The first thing you need to 321 00:16:52,000 --> 00:16:55,440 Speaker 3: do is you need to build guardrails or safeties in there, 322 00:16:55,480 --> 00:16:58,600 Speaker 3: because large language models are essentially just too helpful. If 323 00:16:58,600 --> 00:17:00,480 Speaker 3: they don't know the answer to it, I'll have a go. 324 00:17:00,760 --> 00:17:04,040 Speaker 3: And in an area like Bayrol, you can't have that 325 00:17:04,160 --> 00:17:06,359 Speaker 3: kind of guessing or estimation. What you've got to do 326 00:17:06,400 --> 00:17:08,920 Speaker 3: is give a definitive answer like an expert would, which 327 00:17:08,960 --> 00:17:10,600 Speaker 3: is I'm sorry, I don't know the answer to that, 328 00:17:10,680 --> 00:17:12,240 Speaker 3: or only a little bit more information for you in 329 00:17:12,320 --> 00:17:14,600 Speaker 3: order to get to an answer rather than guess. And 330 00:17:14,680 --> 00:17:18,480 Speaker 3: so actually putting those safeties around it is an engineering task. 331 00:17:18,880 --> 00:17:21,800 Speaker 3: And then the second thing is really curating the data 332 00:17:21,920 --> 00:17:24,240 Speaker 3: that the answers are given from it, and a payroll example, 333 00:17:24,280 --> 00:17:26,679 Speaker 3: that would be we want the core information from the 334 00:17:26,720 --> 00:17:29,680 Speaker 3: agencies involved and not the case law because the case 335 00:17:29,720 --> 00:17:31,920 Speaker 3: law is non determinative, and out of that you then 336 00:17:32,040 --> 00:17:35,080 Speaker 3: get a simple answer, which is you know, to be 337 00:17:35,400 --> 00:17:38,200 Speaker 3: absolutely accurate. We're rolling that into in fact this month 338 00:17:38,240 --> 00:17:40,479 Speaker 3: that's gone into general availability in our customer base. Will 339 00:17:40,520 --> 00:17:42,919 Speaker 3: now be able to ask it any questions it wants to. 340 00:17:43,400 --> 00:17:45,840 Speaker 3: And then you've got the power of tools like generative AI, 341 00:17:45,960 --> 00:17:48,359 Speaker 3: where if you want it, for example, to answer rewards questions, 342 00:17:48,359 --> 00:17:49,760 Speaker 3: you feed at the awards and it can give you 343 00:17:49,800 --> 00:17:52,119 Speaker 3: the answer the next day, which in any kind of 344 00:17:52,119 --> 00:17:54,400 Speaker 3: older version of artificial intelligence would have been a really 345 00:17:54,400 --> 00:17:57,399 Speaker 3: difficult thing to do. You can imagine an insurance in 346 00:17:57,600 --> 00:18:01,840 Speaker 3: law and medicine in all of these spaces. This expert 347 00:18:01,880 --> 00:18:05,840 Speaker 3: era is going to completely change what it is. You know, 348 00:18:05,840 --> 00:18:07,320 Speaker 3: if you're going out there, you can go out there 349 00:18:07,359 --> 00:18:09,760 Speaker 3: better informed, what's a whole of a decision support and 350 00:18:09,800 --> 00:18:11,240 Speaker 3: a lot of the things that would have taken just 351 00:18:11,320 --> 00:18:14,600 Speaker 3: lots of document reading and trying to findances and achieve 352 00:18:14,600 --> 00:18:15,320 Speaker 3: them really quickly. 353 00:18:15,440 --> 00:18:18,760 Speaker 1: The productivity boost that we've been promised from AI. Once 354 00:18:18,960 --> 00:18:24,280 Speaker 1: these systems are used generally across the public sector in 355 00:18:24,359 --> 00:18:26,600 Speaker 1: businesses as well, we will start to see that. But 356 00:18:26,680 --> 00:18:28,920 Speaker 1: one area I know that you're working on as well. 357 00:18:28,960 --> 00:18:31,840 Speaker 1: I've been talking to some of your engineers just about 358 00:18:31,960 --> 00:18:36,680 Speaker 1: how AI is transforming the software development process. You're developing 359 00:18:36,800 --> 00:18:40,720 Speaker 1: applications and platforms for companies all the time, and a 360 00:18:40,720 --> 00:18:42,520 Speaker 1: particular problem we have, we've been a bit slow and 361 00:18:42,600 --> 00:18:46,639 Speaker 1: using some of our organizations to modernize their old systems. 362 00:18:47,000 --> 00:18:50,000 Speaker 1: But to be able to use AI agents actually for 363 00:18:50,760 --> 00:18:55,560 Speaker 1: the software development process, from the actual coding, to project management, 364 00:18:55,640 --> 00:19:02,600 Speaker 1: to business analysis to testing. That is massively improving and 365 00:19:02,640 --> 00:19:05,199 Speaker 1: making more efficient the software development process, isn't it. 366 00:19:05,280 --> 00:19:07,000 Speaker 3: Yeah, it is. And you know, it's one of the 367 00:19:07,000 --> 00:19:11,080 Speaker 3: big ironies that the change in roles in changing what 368 00:19:11,080 --> 00:19:12,880 Speaker 3: you do in your day to day job is possibly 369 00:19:12,920 --> 00:19:15,679 Speaker 3: greatest in the tech sector. As a consequence of it, 370 00:19:15,800 --> 00:19:19,520 Speaker 3: large language models are really good at engineering problems. A 371 00:19:19,520 --> 00:19:23,600 Speaker 3: lot of other organizations are going about improving the productivity 372 00:19:23,640 --> 00:19:26,199 Speaker 3: of their programming in a subtly different way, which is 373 00:19:26,200 --> 00:19:28,679 Speaker 3: giving some of the new generation of tools that enable 374 00:19:29,040 --> 00:19:31,840 Speaker 3: generation of code to their developers. What this is doing 375 00:19:31,920 --> 00:19:35,280 Speaker 3: is it's moving further back up the software development life cycle. 376 00:19:35,560 --> 00:19:39,800 Speaker 3: It is rethinking how you produce the original analysis of 377 00:19:39,840 --> 00:19:42,600 Speaker 3: the system to be built, and then go heavily agent 378 00:19:42,680 --> 00:19:45,240 Speaker 3: oriented for all the rest of the software development life cycle. 379 00:19:45,480 --> 00:19:48,080 Speaker 3: Old systems. If you have the code available to them, 380 00:19:48,320 --> 00:19:51,040 Speaker 3: that code forms the blueprint for what the new system 381 00:19:51,080 --> 00:19:52,960 Speaker 3: needs to do when you want to replace it. Obviously 382 00:19:53,000 --> 00:19:54,399 Speaker 3: there'll be a point where you want to enhance it 383 00:19:54,400 --> 00:19:56,080 Speaker 3: and do a whole lot of modern things, But most 384 00:19:56,119 --> 00:19:58,440 Speaker 3: of these older systems are locked up in a generation 385 00:19:58,600 --> 00:20:00,359 Speaker 3: of code that either there's not a lot a skill 386 00:20:00,400 --> 00:20:02,280 Speaker 3: around it, or you wouldn't want to build and invest 387 00:20:02,280 --> 00:20:04,240 Speaker 3: a whole lot in anyway because it's not so useful. 388 00:20:04,359 --> 00:20:06,520 Speaker 3: The approach that we're talking about came out of an 389 00:20:06,560 --> 00:20:11,040 Speaker 3: opportunity that we have in government in Australia, and the 390 00:20:11,080 --> 00:20:14,919 Speaker 3: team built business analysis agents that go across all of 391 00:20:14,960 --> 00:20:18,280 Speaker 3: the old code and produce the blueprint or the documentation 392 00:20:18,400 --> 00:20:20,760 Speaker 3: for what the new system needs to do based on 393 00:20:21,000 --> 00:20:23,800 Speaker 3: the old code. So the original draft documentation that had 394 00:20:23,840 --> 00:20:27,320 Speaker 3: been produced by the customer represented five years worth of 395 00:20:27,359 --> 00:20:29,800 Speaker 3: business analysis time. The agents were able to do that 396 00:20:30,240 --> 00:20:33,520 Speaker 3: literally in hours by going through this old kind of 397 00:20:33,600 --> 00:20:37,639 Speaker 3: nineties era thirteen hundred form client service system that you 398 00:20:37,640 --> 00:20:40,359 Speaker 3: wouldn't put a lot of investment into in current times. 399 00:20:40,920 --> 00:20:45,520 Speaker 3: Once that blueprint or documentation was available, the team then 400 00:20:45,960 --> 00:20:50,679 Speaker 3: let loose team lead and development agents one of the 401 00:20:50,680 --> 00:20:54,160 Speaker 3: moments where what generator of AI and agents in particular 402 00:20:54,240 --> 00:20:55,800 Speaker 3: can do kind of knock me back in my seat 403 00:20:55,920 --> 00:20:58,359 Speaker 3: for the first time in ages in the tech world, 404 00:20:58,840 --> 00:21:02,720 Speaker 3: was watching a canbam board of these agents just tearing 405 00:21:02,800 --> 00:21:04,600 Speaker 3: through what would have been months and months and months 406 00:21:04,680 --> 00:21:06,960 Speaker 3: and months with a deb effort. And then watching the 407 00:21:07,000 --> 00:21:09,240 Speaker 3: agents talking to each other and celebrating when they did 408 00:21:09,240 --> 00:21:12,680 Speaker 3: a release and congratulating each other for the progress they've 409 00:21:12,720 --> 00:21:16,200 Speaker 3: been making and interacting like a highly automated and highly 410 00:21:16,240 --> 00:21:20,119 Speaker 3: efficient team would, And then seeing the test documentation generated 411 00:21:20,160 --> 00:21:22,679 Speaker 3: and test agents be able to parallel test old system 412 00:21:22,720 --> 00:21:25,159 Speaker 3: news system to be able to determine that the new 413 00:21:25,200 --> 00:21:28,120 Speaker 3: system was delivering the outcomes that needed to wow. Our 414 00:21:28,240 --> 00:21:30,639 Speaker 3: estimate is that in that particular project, we haved our 415 00:21:30,680 --> 00:21:32,880 Speaker 3: original estimate for the dev that needed to be done, 416 00:21:33,160 --> 00:21:35,680 Speaker 3: and over seventy percent of the code of the new 417 00:21:35,680 --> 00:21:40,199 Speaker 3: system was built by agents. Now loads of organizations all 418 00:21:40,240 --> 00:21:42,520 Speaker 3: the way around the world have these older systems that 419 00:21:42,840 --> 00:21:47,000 Speaker 3: replacing is a very, very, very risky task because you know, 420 00:21:47,119 --> 00:21:50,000 Speaker 3: humans going through all the old code and exactly getting 421 00:21:50,040 --> 00:21:53,720 Speaker 3: the modern system right isn't as accurate, unfortunately as actually 422 00:21:53,800 --> 00:21:55,560 Speaker 3: highly automated approaches to it like this. 423 00:21:55,640 --> 00:21:58,560 Speaker 1: If you've got a green screen system, like running a 424 00:21:58,560 --> 00:22:00,640 Speaker 1: bank is something they just don't want it to because 425 00:22:00,680 --> 00:22:01,520 Speaker 1: it's rock solid. 426 00:22:02,560 --> 00:22:05,800 Speaker 3: There are just examples absolutely everywhere you lock in the 427 00:22:05,800 --> 00:22:08,600 Speaker 3: commercial sector and government right the way across the spectrum. 428 00:22:08,800 --> 00:22:12,040 Speaker 3: In order to get that productivity, it wasn't about taking 429 00:22:12,160 --> 00:22:16,920 Speaker 3: one role in that equation, which is like a developer, 430 00:22:17,000 --> 00:22:20,199 Speaker 3: and making it more productive. It was about rethinking the 431 00:22:20,359 --> 00:22:23,080 Speaker 3: entire process to be able to automate the end to 432 00:22:23,240 --> 00:22:27,840 Speaker 3: end process, and so for every other process that exists 433 00:22:27,840 --> 00:22:30,600 Speaker 3: in business where you think that a large language model 434 00:22:30,720 --> 00:22:34,000 Speaker 3: or an AI agent or tool can make a difference, 435 00:22:34,080 --> 00:22:37,000 Speaker 3: what you need to do is do the proof work 436 00:22:37,200 --> 00:22:39,399 Speaker 3: to prove how much of the work can be done 437 00:22:39,760 --> 00:22:43,320 Speaker 3: via the automation, and then completely rethink the process. Not 438 00:22:43,359 --> 00:22:46,320 Speaker 3: just think the individual role, but actually think about how 439 00:22:46,320 --> 00:22:49,040 Speaker 3: that entire process will work differently in your company. And 440 00:22:49,080 --> 00:22:52,119 Speaker 3: it could be legal drafting, it could be giving advice 441 00:22:52,160 --> 00:22:54,399 Speaker 3: to a patient, or perhaps if you wanted to be 442 00:22:54,520 --> 00:22:57,680 Speaker 3: really disruptive, think fully about all the system that happens 443 00:22:57,720 --> 00:23:00,280 Speaker 3: in a ward at a hospital. And I'm sure will 444 00:23:00,280 --> 00:23:01,919 Speaker 3: be parts of the world that are thinking about how 445 00:23:01,960 --> 00:23:04,600 Speaker 3: to revolutionize that because it's a worldwide challenge. There are 446 00:23:04,640 --> 00:23:08,920 Speaker 3: dozens of examples in any kind of space that deals 447 00:23:08,960 --> 00:23:13,240 Speaker 3: heavily with contracts, that deals heavily worth legislation, where you've 448 00:23:13,280 --> 00:23:16,240 Speaker 3: got large amounts of language that are very time consuming 449 00:23:16,280 --> 00:23:18,760 Speaker 3: for people to go through and think about a different outcome. 450 00:23:18,840 --> 00:23:22,359 Speaker 3: And obviously these tools can produce high quality the challenge 451 00:23:22,400 --> 00:23:25,560 Speaker 3: with this, and I'll use the self driving car example 452 00:23:26,240 --> 00:23:29,000 Speaker 3: as a you know, we've been promised self driving cars 453 00:23:29,040 --> 00:23:31,600 Speaker 3: for ages now, right, you know, the revolution where cars 454 00:23:31,600 --> 00:23:33,680 Speaker 3: would be driving themselves every year without people having to 455 00:23:33,720 --> 00:23:35,320 Speaker 3: keep their hands on the wheel. I'm sure it's more 456 00:23:35,359 --> 00:23:37,800 Speaker 3: than a decade old that the conversation about that started. 457 00:23:38,080 --> 00:23:40,840 Speaker 3: The reason it's taken such a long time for it 458 00:23:41,000 --> 00:23:45,520 Speaker 3: to arrive is not because of the day to day 459 00:23:45,560 --> 00:23:49,240 Speaker 3: but because of what happens in harder to product circumstances 460 00:23:49,240 --> 00:23:52,480 Speaker 3: and testing their systems. In the old world of programming, 461 00:23:52,640 --> 00:23:55,240 Speaker 3: you would test for all of the conditions and things 462 00:23:55,280 --> 00:23:59,080 Speaker 3: that happen in a very follow up process kind of way. 463 00:23:59,520 --> 00:24:01,880 Speaker 3: In this the amount of data that you feed an 464 00:24:01,880 --> 00:24:05,159 Speaker 3: expert system and AI system, be it for self driving 465 00:24:05,200 --> 00:24:08,040 Speaker 3: cars or be it for large language models, means that 466 00:24:08,080 --> 00:24:10,720 Speaker 3: you've got to have. The more mission critical, the more 467 00:24:10,720 --> 00:24:15,320 Speaker 3: it deals with lives. And you could treat employment decisions 468 00:24:15,359 --> 00:24:18,200 Speaker 3: as an example of dealing with lives quality employment decisions 469 00:24:18,200 --> 00:24:19,800 Speaker 3: about who you might hire or who you might not, 470 00:24:20,400 --> 00:24:24,480 Speaker 3: decisions about how medical care is, dispense decisions about outcomes 471 00:24:24,520 --> 00:24:28,000 Speaker 3: in court. But how you test the system to ensure 472 00:24:28,040 --> 00:24:31,520 Speaker 3: that in the unusual cases or the edge cases, which 473 00:24:31,560 --> 00:24:34,080 Speaker 3: we call it a programming world, the right outcome has 474 00:24:34,160 --> 00:24:37,720 Speaker 3: arrived at is incredibly complicated. So even though it looks 475 00:24:37,760 --> 00:24:40,280 Speaker 3: like you can get ninety nine percent of the time 476 00:24:40,720 --> 00:24:43,280 Speaker 3: a really expert answer to a problem by using a 477 00:24:43,320 --> 00:24:45,800 Speaker 3: system like this, how you test that to assure that 478 00:24:45,840 --> 00:24:49,600 Speaker 3: you get that in the most important situations is very difficult. 479 00:24:49,760 --> 00:24:52,240 Speaker 3: That's the challenge with rolling their systems out into mission 480 00:24:52,240 --> 00:24:53,159 Speaker 3: critical use cases. 481 00:24:53,200 --> 00:24:58,680 Speaker 1: Well that's why healthcare, insurance, public sector, putting this into 482 00:24:59,280 --> 00:25:03,400 Speaker 1: tax system or social welfare that there's hesitancy, and we've 483 00:25:03,440 --> 00:25:06,680 Speaker 1: been relatively slow to accelerate that because it's those one 484 00:25:06,680 --> 00:25:09,679 Speaker 1: percent of cases, if they go wrong, are really going 485 00:25:09,720 --> 00:25:13,200 Speaker 1: to erode trust in AI systems. So you can understand 486 00:25:13,400 --> 00:25:17,320 Speaker 1: why we're seeing AI agents pop up for various use 487 00:25:17,400 --> 00:25:18,600 Speaker 1: cases but not others. 488 00:25:18,680 --> 00:25:21,280 Speaker 3: Yet, if perhaps some of the effort that was going 489 00:25:21,320 --> 00:25:24,879 Speaker 3: into coding is requiring less effort, how you test the 490 00:25:24,920 --> 00:25:26,840 Speaker 3: output and the system, on the other hand, is becoming 491 00:25:26,880 --> 00:25:30,080 Speaker 3: even more important in this era. And then when you 492 00:25:30,119 --> 00:25:33,760 Speaker 3: get to a situation and we can already see these 493 00:25:33,800 --> 00:25:35,840 Speaker 3: emerging where multiple agents are talking to each other and 494 00:25:35,840 --> 00:25:39,040 Speaker 3: producing group outcomes, you've then got an even more complicated 495 00:25:39,080 --> 00:25:42,480 Speaker 3: set of circumstances to test for because they might not 496 00:25:42,560 --> 00:25:46,040 Speaker 3: even come from the same organization. So while it has 497 00:25:46,200 --> 00:25:49,119 Speaker 3: reduced some of the effort in coding, testing for the 498 00:25:49,119 --> 00:25:51,480 Speaker 3: outcomes and ensuring that you do the design piece right 499 00:25:51,480 --> 00:25:52,960 Speaker 3: has suddenly become more important. Yeah. 500 00:25:53,000 --> 00:25:54,439 Speaker 1: Well, I wanted to ask you about that because the 501 00:25:54,480 --> 00:25:57,879 Speaker 1: obvious question is if you're able to automate seventy percent 502 00:25:57,920 --> 00:26:01,480 Speaker 1: of the code development for an new platform or app, 503 00:26:01,800 --> 00:26:04,600 Speaker 1: what does that mean for your software development workforce? But 504 00:26:04,640 --> 00:26:08,440 Speaker 1: I think you've asked there. It's not necessarily reducing that workforce, 505 00:26:08,480 --> 00:26:10,399 Speaker 1: but they're going to be doing different types of roles. 506 00:26:10,640 --> 00:26:12,480 Speaker 3: Yeah, and I think you're going to see this. My 507 00:26:12,600 --> 00:26:15,120 Speaker 3: role has changed on me absolutely right the way across 508 00:26:15,160 --> 00:26:17,879 Speaker 3: the spectrum of adoption of generator b AI. If we 509 00:26:17,920 --> 00:26:20,359 Speaker 3: go one step further. What's going to happen to employment. 510 00:26:20,400 --> 00:26:23,320 Speaker 3: That's the hard question. Will it completely change the landscape 511 00:26:23,359 --> 00:26:26,480 Speaker 3: of what happens in the technology world. Yes. Is there 512 00:26:26,480 --> 00:26:29,040 Speaker 3: more technology work to do and more work to do 513 00:26:29,119 --> 00:26:31,880 Speaker 3: to implement these kind of systems for companies? Yes. Will 514 00:26:31,880 --> 00:26:34,359 Speaker 3: it mean a reduction in the tech sector at the moment, 515 00:26:34,440 --> 00:26:36,439 Speaker 3: I think that's highly unlikely. I think the need for 516 00:26:36,480 --> 00:26:38,840 Speaker 3: this kind of productivity is going to be critical to 517 00:26:38,840 --> 00:26:42,800 Speaker 3: be globally competitive. What it means for broader employment is 518 00:26:42,840 --> 00:26:45,600 Speaker 3: a more complicated one. And you know, economists have been 519 00:26:45,600 --> 00:26:48,879 Speaker 3: writing about this actually since pre the GENAIRA, since the 520 00:26:48,920 --> 00:26:52,720 Speaker 3: AI era. They don't agree. I don't agree in the slightest. 521 00:26:52,760 --> 00:26:55,960 Speaker 3: There are some that predict mass in equity and a 522 00:26:56,040 --> 00:27:00,000 Speaker 3: hollowing out of some job types, and certainly you can 523 00:27:00,080 --> 00:27:03,280 Speaker 3: see some of that trend at the moment. We've been 524 00:27:03,359 --> 00:27:06,840 Speaker 3: through big revolutions like this before the Industrial Revolution and 525 00:27:06,880 --> 00:27:10,159 Speaker 3: other ones, and new kinds of jobs emerged and new 526 00:27:10,240 --> 00:27:12,439 Speaker 3: kinds of opportunity emerged out the other side of it, 527 00:27:12,880 --> 00:27:16,760 Speaker 3: and then opportunity rebalanced in and around it. So at 528 00:27:16,760 --> 00:27:19,280 Speaker 3: the moment, I'd prefer to stay on the side of 529 00:27:19,400 --> 00:27:23,640 Speaker 3: tech optimist in that as long as everybody stays focused 530 00:27:23,640 --> 00:27:26,760 Speaker 3: on the beneficial applications of it, then I think will 531 00:27:26,760 --> 00:27:31,840 Speaker 3: make progress. But the short term impact could be quite dramatic, 532 00:27:32,640 --> 00:27:35,000 Speaker 3: and it's certainly it's going to challenge what people learn 533 00:27:35,320 --> 00:27:39,240 Speaker 3: in school and learn and university. I know the university 534 00:27:39,280 --> 00:27:42,000 Speaker 3: is a hugely challenged by this just from evaluating student performance. 535 00:27:42,480 --> 00:27:44,520 Speaker 3: How can you tell what's AI written and what student 536 00:27:44,760 --> 00:27:48,280 Speaker 3: It's getting increasingly difficult what people must be learning to 537 00:27:48,400 --> 00:27:51,520 Speaker 3: use these alongside their university learning in order to be 538 00:27:51,520 --> 00:27:54,439 Speaker 3: productive in the workplace using this kind of augmentation. And 539 00:27:54,480 --> 00:27:58,240 Speaker 3: it's changing really rapidly, and so quite what it means 540 00:27:58,800 --> 00:28:01,119 Speaker 3: for the future is very very difficult to protect them. 541 00:28:01,160 --> 00:28:04,159 Speaker 3: I prefer an economist or a sociologist or somebody like 542 00:28:04,200 --> 00:28:06,840 Speaker 3: that was interviewed rather than I tried to guess that. 543 00:28:07,119 --> 00:28:09,920 Speaker 3: On the other hand, I do think that we can't 544 00:28:10,680 --> 00:28:14,119 Speaker 3: avoid the fact that all of the global organizations that 545 00:28:14,440 --> 00:28:16,920 Speaker 3: are competing in our market, are competing for work, are 546 00:28:16,920 --> 00:28:19,879 Speaker 3: adopting it, and therefore we must evolve in order to 547 00:28:19,920 --> 00:28:22,040 Speaker 3: be competitive in that global stage. 548 00:28:22,080 --> 00:28:24,800 Speaker 1: We do have a productivity problem. It's very competitive. We 549 00:28:24,840 --> 00:28:28,280 Speaker 1: do need to adopt AI. We've just seen the government's 550 00:28:28,280 --> 00:28:31,080 Speaker 1: AI strategy for the country. We were the last in 551 00:28:31,080 --> 00:28:34,320 Speaker 1: the OECD to publish one. It's had a fair bit 552 00:28:34,359 --> 00:28:36,800 Speaker 1: of criticism. But as the leader of one of the 553 00:28:37,960 --> 00:28:41,160 Speaker 1: biggest IT company in New Zealand, you think we've got 554 00:28:41,200 --> 00:28:44,840 Speaker 1: the settings right. Is there enough urgency and momentum in 555 00:28:44,880 --> 00:28:48,000 Speaker 1: what the government has outlined to sort of get us here, 556 00:28:48,080 --> 00:28:51,800 Speaker 1: to get us those productivity gains and get AI competently 557 00:28:51,880 --> 00:28:54,520 Speaker 1: and ethically used across New Zealand. 558 00:28:56,360 --> 00:29:01,680 Speaker 3: I think asking government to do this by itself as 559 00:29:01,680 --> 00:29:05,160 Speaker 3: opposed to in cooperation with industry is you know, it's 560 00:29:05,200 --> 00:29:07,480 Speaker 3: a difficult ask. Government has its own set of priorities 561 00:29:07,520 --> 00:29:10,480 Speaker 3: where it will be focusing on where each of the 562 00:29:10,520 --> 00:29:16,440 Speaker 3: agencies that's in government can effectively adopt adopt the technology. 563 00:29:17,120 --> 00:29:19,280 Speaker 3: What I would what I do think it will be 564 00:29:19,320 --> 00:29:22,800 Speaker 3: worth us touching on is actually the infrastructure needed to 565 00:29:22,800 --> 00:29:25,080 Speaker 3: support AI, because that's the bit of the Like everybody's 566 00:29:25,080 --> 00:29:28,240 Speaker 3: focusing on the ethical conversations. I think there are governments 567 00:29:28,280 --> 00:29:30,080 Speaker 3: around the work that there are a lot of government 568 00:29:30,080 --> 00:29:32,320 Speaker 3: settings that have been put out of that, ranging from 569 00:29:33,240 --> 00:29:37,720 Speaker 3: you know the US UK examples through to the EU examples, 570 00:29:37,760 --> 00:29:42,720 Speaker 3: through to the work that Australia is doing and some 571 00:29:42,800 --> 00:29:44,640 Speaker 3: of the ethical pieces that need to be tackled there. 572 00:29:44,640 --> 00:29:46,800 Speaker 3: But let's talk about the infrastructure that underpends AI and 573 00:29:46,840 --> 00:29:49,960 Speaker 3: the huge consequences of that. Where New Zealand has some 574 00:29:50,040 --> 00:29:54,160 Speaker 3: distinct advantages and it has some really important decisions that 575 00:29:54,160 --> 00:29:59,760 Speaker 3: it's going to have to make going forward. So we've 576 00:29:59,840 --> 00:30:02,080 Speaker 3: to talked about the large language models. What we haven't 577 00:30:02,120 --> 00:30:05,240 Speaker 3: talked about is the infrastructure needed in order to produce 578 00:30:05,280 --> 00:30:12,960 Speaker 3: those outcomes. And so AI is the GENAI answers rely 579 00:30:13,120 --> 00:30:16,880 Speaker 3: on two things happening. They rely on the large language 580 00:30:16,880 --> 00:30:20,240 Speaker 3: models being trained. So that is a very very very 581 00:30:21,960 --> 00:30:26,200 Speaker 3: compute intensor process that uses GPUs rather than the old 582 00:30:26,280 --> 00:30:28,960 Speaker 3: CPU world of the computer world. That's why you've seen 583 00:30:29,000 --> 00:30:31,120 Speaker 3: in Video become one of the most valuable companies in 584 00:30:31,160 --> 00:30:33,880 Speaker 3: the world. And you've seen AMD put a really strong 585 00:30:33,960 --> 00:30:37,360 Speaker 3: kind of parallel response alongside that about chipset capability that 586 00:30:37,360 --> 00:30:42,680 Speaker 3: it's got in the US. That means that the data 587 00:30:42,680 --> 00:30:46,000 Speaker 3: center market has been lit on fire, as has that 588 00:30:46,120 --> 00:30:50,200 Speaker 3: of providing GPUs to support both training to the two pieces, 589 00:30:50,240 --> 00:30:52,440 Speaker 3: there's the training of the large language model, that inference, 590 00:30:52,600 --> 00:30:54,480 Speaker 3: which is a jargon term. I'm going to use because 591 00:30:54,480 --> 00:30:56,600 Speaker 3: I won't be able to help myself. Is when you 592 00:30:56,680 --> 00:30:58,560 Speaker 3: send a query to a large language model and get 593 00:30:58,560 --> 00:31:02,840 Speaker 3: an answer back. Now, to train, I think it was 594 00:31:03,040 --> 00:31:08,000 Speaker 3: GPT three point five two point eight million hours of 595 00:31:08,360 --> 00:31:12,480 Speaker 3: GPU time per training run. You've got to think cards 596 00:31:12,520 --> 00:31:15,640 Speaker 3: sitting in a server and two point eight million hours 597 00:31:15,720 --> 00:31:19,880 Speaker 3: worth of one of these expensive Nvidia cards with all 598 00:31:19,880 --> 00:31:22,080 Speaker 3: the heat and power associated with that. So in the 599 00:31:22,160 --> 00:31:24,360 Speaker 3: US they're down to less than four percent data center 600 00:31:24,360 --> 00:31:29,000 Speaker 3: space available. They're firing up old nuclear reactors in order 601 00:31:29,040 --> 00:31:33,360 Speaker 3: to be able to fuel the demand for training activity. 602 00:31:33,520 --> 00:31:35,560 Speaker 3: The inference side of it is still a big demand, 603 00:31:35,560 --> 00:31:40,560 Speaker 3: and it's rapidly accelerating, and only a few parts of 604 00:31:40,560 --> 00:31:43,719 Speaker 3: the world are able to take those advanced chipsets because 605 00:31:43,720 --> 00:31:46,040 Speaker 3: of the legal restrictions that have been put in place 606 00:31:46,680 --> 00:31:49,960 Speaker 3: between the US and China. Now, the interesting part for 607 00:31:50,680 --> 00:31:56,280 Speaker 3: where we are is Japan, South Korea, Australia and New 608 00:31:56,320 --> 00:31:59,360 Speaker 3: Zealand are the only countries that can actually receive these 609 00:31:59,400 --> 00:32:02,200 Speaker 3: chipsets this part of the world. So all of the 610 00:32:02,440 --> 00:32:07,480 Speaker 3: inference that is done for all of our part of 611 00:32:07,520 --> 00:32:09,120 Speaker 3: the world that isn't going to go to the US 612 00:32:09,360 --> 00:32:13,000 Speaker 3: can only land in one of those countries, but there 613 00:32:13,040 --> 00:32:16,880 Speaker 3: needs to be sufficient power to fuel that, so there's 614 00:32:16,920 --> 00:32:19,560 Speaker 3: going to be potentially a huge crunch on power. You 615 00:32:19,600 --> 00:32:21,479 Speaker 3: can already see it starting to turn up in some 616 00:32:21,520 --> 00:32:25,200 Speaker 3: parts of Australia. Long term forward planning is needed to 617 00:32:25,440 --> 00:32:28,240 Speaker 3: ensure the powers and of course we've got this huge 618 00:32:28,280 --> 00:32:31,240 Speaker 3: advantage in New Zealand of renewable power if we invest 619 00:32:31,320 --> 00:32:33,240 Speaker 3: enough in the infrastructure and the transmission in order to 620 00:32:33,280 --> 00:32:36,720 Speaker 3: be able to do it. But we don't have a 621 00:32:36,720 --> 00:32:38,920 Speaker 3: lot of that processing happening here at most of what 622 00:32:39,000 --> 00:32:41,280 Speaker 3: you do when you interact with one of these products, 623 00:32:41,280 --> 00:32:42,960 Speaker 3: we'll go to another part of the world to be processed. 624 00:32:43,520 --> 00:32:45,440 Speaker 3: So do we need the infrastructure for it here? I'd 625 00:32:45,440 --> 00:32:48,520 Speaker 3: say from a resilience perspective, we do is the forward 626 00:32:48,560 --> 00:32:52,560 Speaker 3: investment in power and data center infrastructure and GPU deployment 627 00:32:52,600 --> 00:32:55,160 Speaker 3: can happen to do it? That's very uncertain and it's 628 00:32:55,200 --> 00:32:58,720 Speaker 3: not being discussed enough as everybody's focusing further up the 629 00:32:58,720 --> 00:33:02,080 Speaker 3: stack on how to make their individual organization or productive. 630 00:33:02,600 --> 00:33:05,560 Speaker 3: So I think, I think how we have the and 631 00:33:05,600 --> 00:33:09,160 Speaker 3: then of course you know we well, there's a lot 632 00:33:09,160 --> 00:33:10,800 Speaker 3: of other places we could go from there. I won't 633 00:33:11,400 --> 00:33:13,920 Speaker 3: I won't go down too many rabbit holes. But I 634 00:33:13,960 --> 00:33:16,080 Speaker 3: don't hear that being talked about anywhere. No, it's not 635 00:33:16,440 --> 00:33:18,320 Speaker 3: just sort of this assumption that it will be there 636 00:33:18,560 --> 00:33:22,440 Speaker 3: and it won't because it's it's as you know, hundreds 637 00:33:22,480 --> 00:33:25,280 Speaker 3: of millions or billions worth of investment needed in order 638 00:33:25,320 --> 00:33:27,400 Speaker 3: to support that kind of process and capability. 639 00:33:27,960 --> 00:33:31,480 Speaker 1: Yeah, and is it I mean, Datacom is investing in 640 00:33:31,520 --> 00:33:37,040 Speaker 1: these GPUs. You know, you've got extensive data center infrastructure, 641 00:33:37,280 --> 00:33:41,760 Speaker 1: so you're upgrading it for for the A centric applications. 642 00:33:42,360 --> 00:33:45,280 Speaker 1: I mean, is it completely unrealistic for us to ever 643 00:33:45,560 --> 00:33:48,680 Speaker 1: as a nation develop our own LLM or at least 644 00:33:48,920 --> 00:33:54,040 Speaker 1: adapt and open source LLM. The Swiss have just created one, 645 00:33:54,360 --> 00:33:56,760 Speaker 1: spent a lot of money developing it, they've open sourced 646 00:33:56,800 --> 00:33:59,360 Speaker 1: it under an Apache license. So we could we take that, 647 00:34:00,120 --> 00:34:03,920 Speaker 1: train it or adapt it for our uses at reasonable 648 00:34:03,920 --> 00:34:06,280 Speaker 1: cost or is it just prohibitively expensive. 649 00:34:08,320 --> 00:34:12,080 Speaker 3: It's probably worth talking about what's happening in other parts 650 00:34:12,080 --> 00:34:15,680 Speaker 3: of the world. It's a really really interesting question because 651 00:34:15,680 --> 00:34:17,840 Speaker 3: there are multiple stages in what you do when you 652 00:34:17,840 --> 00:34:23,239 Speaker 3: get an LM to produce answers. Obviously, you know there's 653 00:34:23,280 --> 00:34:25,600 Speaker 3: a need for an LIM that can speak today in 654 00:34:26,080 --> 00:34:29,360 Speaker 3: New Zealand, and I know there's lots of pockets of 655 00:34:29,400 --> 00:34:32,120 Speaker 3: activity happening there. I'm positive that all the pockets of 656 00:34:32,520 --> 00:34:35,480 Speaker 3: activity need to cooperate for that outcome because it will 657 00:34:35,480 --> 00:34:38,239 Speaker 3: be one that will require tremendous investment and effort in 658 00:34:38,320 --> 00:34:42,440 Speaker 3: order to deliver the outcomes. It is much more likely 659 00:34:43,000 --> 00:34:45,239 Speaker 3: that we will do a lot of fine tuning of 660 00:34:45,280 --> 00:34:47,440 Speaker 3: a large language model, which is a stage that happens 661 00:34:47,440 --> 00:34:52,120 Speaker 3: after training, because investing in the training also requires massive 662 00:34:52,160 --> 00:34:53,879 Speaker 3: amounts of data in order to do it in order 663 00:34:53,920 --> 00:34:56,560 Speaker 3: to make the LM able to deal with more difficult questions. 664 00:34:57,320 --> 00:35:03,879 Speaker 3: So the fine tuning piece is in essence, selecting where 665 00:35:03,880 --> 00:35:06,960 Speaker 3: it is expert and what information it relies on, and 666 00:35:07,040 --> 00:35:09,920 Speaker 3: waiting the answers. So some parts of the world that 667 00:35:10,040 --> 00:35:15,040 Speaker 3: perhaps have different views on history are doing it at 668 00:35:15,040 --> 00:35:17,160 Speaker 3: the fine tuning stage in order to say that in 669 00:35:17,200 --> 00:35:19,560 Speaker 3: our country you want history to be represented this way 670 00:35:19,600 --> 00:35:24,360 Speaker 3: when the large language model because it's answer and you 671 00:35:24,480 --> 00:35:28,000 Speaker 3: then get into a whole bunch of really really really 672 00:35:28,040 --> 00:35:30,920 Speaker 3: interesting questions that you could you know, if you get 673 00:35:30,920 --> 00:35:32,560 Speaker 3: some of the greatest experts of this in the world 674 00:35:32,560 --> 00:35:35,320 Speaker 3: around the table. If you think about the interaction between 675 00:35:35,360 --> 00:35:40,279 Speaker 3: language and worldview. Is the fact that the majority of 676 00:35:40,280 --> 00:35:42,680 Speaker 3: these are trained in English first going to result in 677 00:35:42,680 --> 00:35:44,799 Speaker 3: a limited worldview. Can you deal with that at fine 678 00:35:44,840 --> 00:35:52,360 Speaker 3: tuning level? Which version of history, what answers is it 679 00:35:52,400 --> 00:35:55,600 Speaker 3: going to give to critical questions of history or belief? 680 00:35:57,200 --> 00:36:00,359 Speaker 3: Is that intrinsic to the training that it's received, or 681 00:36:00,400 --> 00:36:02,560 Speaker 3: can that be dealt with at fine tuning. But there's 682 00:36:02,640 --> 00:36:06,080 Speaker 3: no doubt in my mind that the place that young 683 00:36:06,120 --> 00:36:09,640 Speaker 3: people go to when they're at high school or university 684 00:36:09,719 --> 00:36:12,640 Speaker 3: is going to become large language models very rapidly instead 685 00:36:12,640 --> 00:36:15,399 Speaker 3: of the more traditional search world. And so actually it's 686 00:36:15,480 --> 00:36:19,000 Speaker 3: really important to understand what biases the language models have 687 00:36:19,040 --> 00:36:23,480 Speaker 3: when they answer those questions. When people delve into looking 688 00:36:23,800 --> 00:36:26,319 Speaker 3: for help in their life, will they ask at large 689 00:36:26,360 --> 00:36:28,480 Speaker 3: language model that question? Is it safe to give answers 690 00:36:28,480 --> 00:36:31,680 Speaker 3: to that? What will it say about the history of Altro? 691 00:36:32,080 --> 00:36:34,160 Speaker 3: What will it say about the history of other countries 692 00:36:34,200 --> 00:36:35,879 Speaker 3: in the world. And we need to take a view 693 00:36:35,920 --> 00:36:38,800 Speaker 3: on that. And so I don't see that debate happening 694 00:36:38,920 --> 00:36:42,799 Speaker 3: enough either. And I think, you know, we're drifting towards 695 00:36:43,040 --> 00:36:45,120 Speaker 3: what does it mean. I don't think we'll have to 696 00:36:45,160 --> 00:36:47,319 Speaker 3: train them from scratch in order to do it, and 697 00:36:47,360 --> 00:36:50,200 Speaker 3: some of the open source ones are now immensely capable, 698 00:36:50,280 --> 00:36:52,360 Speaker 3: as is the ability to fine tune the larger ones. 699 00:36:52,520 --> 00:36:53,560 Speaker 3: How far do you want to go into this? 700 00:36:53,719 --> 00:36:55,640 Speaker 1: So clearly you know what you're saying is the focus 701 00:36:55,719 --> 00:36:59,360 Speaker 1: on infrastructure. We've got to get that right, not just 702 00:36:59,400 --> 00:37:01,600 Speaker 1: the data seene infrastructure, and it's great to see more 703 00:37:01,840 --> 00:37:04,120 Speaker 1: a lot of investment going into that space in New Zealand, 704 00:37:04,200 --> 00:37:07,520 Speaker 1: but the power infrastructure, the ethics and the fine tuning 705 00:37:07,560 --> 00:37:10,320 Speaker 1: of this how we go about that's really important. But 706 00:37:10,880 --> 00:37:14,359 Speaker 1: ultimately we can to have some sovereignty I guess over 707 00:37:15,120 --> 00:37:17,919 Speaker 1: AIO or be able to steer our destiny rather than 708 00:37:18,120 --> 00:37:22,080 Speaker 1: just rely on open AI and anthropic to provide these great, 709 00:37:22,239 --> 00:37:26,160 Speaker 1: large language models that don't necessarily represent our worldview or history. 710 00:37:26,280 --> 00:37:28,279 Speaker 1: I guess we've got to get really deliberate about that 711 00:37:28,320 --> 00:37:29,960 Speaker 1: as a nation, don't we. And that's maybe where the 712 00:37:30,000 --> 00:37:33,279 Speaker 1: government can play a role in bringing private sector, academia 713 00:37:33,560 --> 00:37:35,080 Speaker 1: and the government together to do that. 714 00:37:35,280 --> 00:37:38,280 Speaker 3: Yeah, it can, and look, different governments around the world 715 00:37:38,560 --> 00:37:41,799 Speaker 3: have approached it in loads of different ways. We are 716 00:37:42,239 --> 00:37:46,880 Speaker 3: a small country with only five million odd people in 717 00:37:46,920 --> 00:37:50,600 Speaker 3: it and a very finite amount of money that we 718 00:37:50,640 --> 00:37:52,840 Speaker 3: can spend, so we have to be smart. We probably 719 00:37:52,880 --> 00:37:56,160 Speaker 3: need to choose very carefully who were tightly allied with 720 00:37:56,239 --> 00:37:58,680 Speaker 3: and where we can cooperate and not reinvent stuff that 721 00:37:58,680 --> 00:38:01,080 Speaker 3: we don't need to reinvent, so that we can really 722 00:38:01,120 --> 00:38:03,759 Speaker 3: focus on the core of what making progress looks like. 723 00:38:03,880 --> 00:38:07,760 Speaker 3: And different governments have cooperated public private very very differently 724 00:38:07,800 --> 00:38:09,719 Speaker 3: in different ways in order to move that forward. You know, 725 00:38:09,760 --> 00:38:12,640 Speaker 3: if you look at very large, wealthy countries, what they 726 00:38:12,680 --> 00:38:15,480 Speaker 3: are able to invest in maybe different to what we 727 00:38:15,520 --> 00:38:17,360 Speaker 3: can afford to do here in New Zealand. So the 728 00:38:17,400 --> 00:38:21,000 Speaker 3: decisions we make dealing with scarce capacity to invest in 729 00:38:21,120 --> 00:38:23,640 Speaker 3: R and D have to be really smart and really 730 00:38:23,680 --> 00:38:25,600 Speaker 3: focused on the best outcomes for New Zealand and the 731 00:38:25,600 --> 00:38:27,080 Speaker 3: best outcomes for the people here. 732 00:38:27,440 --> 00:38:32,440 Speaker 1: Just finally, Greg eighteen years as CEO of data Com 733 00:38:32,640 --> 00:38:36,160 Speaker 1: expansion across the Tasman employing over five thousand people. How 734 00:38:36,200 --> 00:38:39,960 Speaker 1: do you stay motivated inspired to lead a big company, 735 00:38:39,960 --> 00:38:43,799 Speaker 1: a huge team, with such change going on? How do 736 00:38:43,800 --> 00:38:47,160 Speaker 1: you keep fresh in that role nearly two decades into it. 737 00:38:47,200 --> 00:38:49,279 Speaker 3: I'm the child of two Tetchers, right, so I was 738 00:38:49,280 --> 00:38:52,640 Speaker 3: brought up on learning and brought up encouraged. I was 739 00:38:52,680 --> 00:38:54,520 Speaker 3: really lucky I was in an environment I was encouraged 740 00:38:54,520 --> 00:38:57,879 Speaker 3: to be curious about things. That curiosity is still there, 741 00:38:58,000 --> 00:39:02,680 Speaker 3: that interest in I started building Lego when I was 742 00:39:02,760 --> 00:39:05,080 Speaker 3: very little, and then found computers and found that what 743 00:39:05,120 --> 00:39:07,640 Speaker 3: you could build on those was whatever I could imagine 744 00:39:07,640 --> 00:39:10,440 Speaker 3: to some extent or other. And so that interest in 745 00:39:10,520 --> 00:39:13,319 Speaker 3: technology and how it applies there and then if I'm 746 00:39:13,320 --> 00:39:15,040 Speaker 3: going to add a second piece to it, which is 747 00:39:15,080 --> 00:39:18,880 Speaker 3: sort of a sense of responsibility I touched very earlier, 748 00:39:19,239 --> 00:39:23,560 Speaker 3: much earlier in our conversation on what a unique position 749 00:39:23,600 --> 00:39:25,960 Speaker 3: in the New Zealand and Australian market and the fact 750 00:39:26,080 --> 00:39:29,560 Speaker 3: that we need voices that worry about how technology is 751 00:39:29,600 --> 00:39:32,200 Speaker 3: implemented in our part of the world. And so I 752 00:39:32,280 --> 00:39:35,440 Speaker 3: think we need organizations that are independent and don't just 753 00:39:35,520 --> 00:39:39,759 Speaker 3: result in this displacement of everything that is spent in 754 00:39:39,800 --> 00:39:42,520 Speaker 3: New Zealand to global forces, and instead we need organizations 755 00:39:42,520 --> 00:39:43,840 Speaker 3: that are based in this part of the world to 756 00:39:43,880 --> 00:39:49,640 Speaker 3: have a voice that ensure that New Zealand's and Australia's 757 00:39:50,239 --> 00:39:53,400 Speaker 3: adoption of technology goes in a way that's good for 758 00:39:53,440 --> 00:39:55,799 Speaker 3: the country we live in and the part of. 759 00:39:55,800 --> 00:39:58,120 Speaker 1: The world we work and it's been a very successful 760 00:39:58,160 --> 00:40:04,000 Speaker 1: formula and sucty. Congratulations on that massive milestone and here's 761 00:40:04,360 --> 00:40:07,160 Speaker 1: too many more decades fulfilling that role in New Zealand. 762 00:40:07,160 --> 00:40:08,960 Speaker 1: Thanks so much for coming on the Business of Tech. 763 00:40:09,000 --> 00:40:10,759 Speaker 3: Thanks so much better, really nice to see you, and 764 00:40:11,000 --> 00:40:11,840 Speaker 3: thanks for conversation. 765 00:40:15,440 --> 00:40:17,520 Speaker 1: That's all for this edition of the Business of Tech. 766 00:40:17,560 --> 00:40:19,920 Speaker 1: My thanks to Greg Davidson for joining me on the 767 00:40:19,960 --> 00:40:24,480 Speaker 1: show and sharing his perspective on Datacom's remarkable sixty year 768 00:40:24,600 --> 00:40:27,759 Speaker 1: journey and what the future holds. The thing that really 769 00:40:27,800 --> 00:40:32,719 Speaker 1: strikes me about Datacom is what Greg said about John Holsworth, 770 00:40:33,000 --> 00:40:36,040 Speaker 1: the early shareholder in Datacom who went on to be 771 00:40:36,360 --> 00:40:39,760 Speaker 1: its executive chairman of the company for around twenty years 772 00:40:39,800 --> 00:40:43,480 Speaker 1: from nineteen ninety and whose family are still a major 773 00:40:43,560 --> 00:40:44,960 Speaker 1: shareholder in the company. 774 00:40:45,600 --> 00:40:47,720 Speaker 2: It was really that relentless. 775 00:40:47,160 --> 00:40:51,839 Speaker 1: Focus on what the customer needed, rather than embracing tech 776 00:40:51,960 --> 00:40:56,759 Speaker 1: fads of the day, that kept Datacom competitive as other 777 00:40:56,800 --> 00:41:00,000 Speaker 1: players emerged in IT services in New Zealand. 778 00:41:00,640 --> 00:41:02,040 Speaker 2: Greg made pretty clear. 779 00:41:01,800 --> 00:41:05,120 Speaker 1: That Datacom's success has always come from that willingness to 780 00:41:05,200 --> 00:41:09,720 Speaker 1: listen to customers but also remaining independent in a field 781 00:41:09,800 --> 00:41:14,880 Speaker 1: of global giants. Datacom isn't the flashiest company around. It 782 00:41:14,920 --> 00:41:20,240 Speaker 1: doesn't like talking about itself particularly, but it's kept loyal 783 00:41:20,360 --> 00:41:25,120 Speaker 1: customers throughout the decades and kept the trust in its 784 00:41:25,160 --> 00:41:28,560 Speaker 1: products and services, which has really paid off in terms 785 00:41:28,600 --> 00:41:32,200 Speaker 1: of its growth and its reach in the region. If 786 00:41:32,239 --> 00:41:35,480 Speaker 1: AI is going to boost our national productivity, which we 787 00:41:35,560 --> 00:41:39,120 Speaker 1: really need, we actually need Datacom to really get it right, 788 00:41:39,440 --> 00:41:41,399 Speaker 1: given that it's such a big player in the IT 789 00:41:41,680 --> 00:41:42,799 Speaker 1: services market here. 790 00:41:43,200 --> 00:41:45,600 Speaker 2: So it was really interesting to hear Greg. 791 00:41:45,360 --> 00:41:48,880 Speaker 1: Talk about the approach to AI and how Datacom is 792 00:41:48,960 --> 00:41:53,560 Speaker 1: changing itself to embrace this new technology. If you liked 793 00:41:53,600 --> 00:41:56,240 Speaker 1: the episode of The Business of Tech, leave a review 794 00:41:56,280 --> 00:41:59,359 Speaker 1: and share the podcast. It's on iHeartRadio and in your 795 00:41:59,360 --> 00:42:02,320 Speaker 1: favorite pod cast app. Check out the show notes in 796 00:42:02,440 --> 00:42:05,360 Speaker 1: my weekly tech reading list over at Businessdesk dot co 797 00:42:05,600 --> 00:42:06,080 Speaker 1: dot Nz. 798 00:42:06,280 --> 00:42:08,080 Speaker 2: You'll find them in the podcast section. 799 00:42:09,040 --> 00:42:11,760 Speaker 1: Next week, I catch you up with a Kiwi xpat 800 00:42:11,840 --> 00:42:15,040 Speaker 1: who is one of the leaders of Britain's efforts to 801 00:42:15,120 --> 00:42:18,640 Speaker 1: build quantum computers. He tells me what the tech has 802 00:42:18,640 --> 00:42:21,120 Speaker 1: in store for us and just where we are in 803 00:42:21,200 --> 00:42:23,640 Speaker 1: the quantum race. That's next Thursday. 804 00:42:23,880 --> 00:42:24,680 Speaker 2: I'll catch you then