← The Next Innovation
Loading summary
Transcript40 lines
- [00:03]
A
Welcome to the next innovation. I'm Jennifer Strong. We're in Austin, Texas for South by Southwest, where this year at Ireland House, we're exploring how Irish companies are using different forms of AI to gain efficiency and more. Perry Cadaster is the Chief Communications Officer of Nearform, a software company that develops tailored AI solutions for clients around the world.
- [00:24]
B
What we're doing is called AI Native engineering, where we're taking ostensibly spec driven development and basically developing agents that can execute a lot of the manual coding that heretofore was obviously done by people. And what that enables is massive, massive efficiency gain.
- [00:44]
A
And Brenda Jordan is the CEO of SOBE Analytics, a company that uses AI to help other companies make smarter financial decisions.
- [00:52]
C
What we've done in Ask SOBE now is, and I would say that we've got 80% AI, but the 20% is human intelligence. And that has got to be filled by people who have expertise and who actually are able to communicate with the business owners and communicate with users.
- [01:11]
A
Let's take a listen. All right, thank you, everybody. Welcome. Appreciate you joining us this morning. All right, so we're here to explore Ireland's rise as a global tech powerhouse as seen through the eyes of these two executives who are living that story from the inside. Right. If you're wondering what I mean by all that, the rise of Ireland, there is quite a bit of data I could point to right, like that. The IMF ranks Ireland first globally for AI readiness in the workforce. The country also ranks first for ICT services exports. And just overall, the country is investing and experimenting and innovating right at the cutting edge, from fintech to enterprise software to global governance. All right, so ladies, speaking of transitions, as I Transitioned to covering AI in 2016 you2 both have also transitioned to AI in your jobs. It's not something you always did. What was that experience like? How did that come to be?
- [02:14]
C
So my background is I spent over 20 years running my own accountancy practice back in Ireland. One of the things that we focused very much on, we did all the usual stuff that accountants do. But one of the things that I myself was very passionate about was delivering growth advice to business owners. And as that part of our business started to grow in popularity, I found more and more and more of my time was being spent in spreadsheets, Doing spreadsheets. Five, six years ago, I went to market looking for a solution that would automate that part of the technology. But what I actually really wanted was I didn't just want the workflow to be automated. I always dreamed and it really was a dream of being able to deliver prescriptive guidance to business owners. So to be able to take information, basically look and say, this is what happens, this is why it happened, and actually this is what you need to do about it. Okay? And it was really the advent of ChatGPT a couple of years ago that actually, you know, I had a lot of people saying to me, oh, that's not the end of, you know, sobe. And I was saying, it's the beginning, because this is actually how we are actually going to deliver this guidance piece that we want. So that really has brought us on a journey. And we are releasing this new product that we have been working on really hard, I could say, for the last six months. But it's more like the last three years, I would say, in reality, where we actually, we looked at the product, we looked at the market, and we realized, okay, you know what, technology is great, AI is great, but we also need that bit of human intelligence or that human touch as well. And so bringing us into AI, that's actually what I guess my journey has been. And it's been a very, I guess, operational type of approach to AI. It's been looking and saying, what is the problem? How do we solve it? And actually realizing actually, you know, something technology and particular AI, it can actually deal with parts of this journey in a way I could only. I couldn't actually even have imagined was beyond the imagination. And so that's been incredibly an incredible journey for me and my team.
- [04:41]
A
Perry, how about you?
- [04:42]
B
I think we've all been using AI longer than we realize, if we're going to be honest. I mean, the level of hyper personalization I have from Amazon is not new. And just knowing now that it's from all the data and the algorithms they have feels both. I know I'm getting marketed to, but it still works every time. So. Thank you, Jeff Bezos. But, you know, beyond that, I'd say ChatGPT was also kind of my gateway drug into AI, so to speak, just because it was easily accessible. And I do believe in the shiny toy kind of use factor of if you give people, you know, technical or not, something that's engaging to use and see value from very quickly, it makes them much more open to adopting that technology. And I think ChatGPT was kind of that shiny toy, if you will, for. For generative AI. I'd say in my role, doing marketing, communications, a lot of what I learned about AI is just from talking to other domain experts in my network from, you know, Also in media and so forth. What do they recommend? Because if you're not an engineer, where do you get started? So Jasper is the tool that I use a lot. It does everything from photo editing to drafting social media posts to spitting out your whole content calendar for a year now do I take everything from that and copy paste? No, but a lot of times it will give me inputs and ideas for what I'm working on. I'd say if I take a step back though, at Nearform, I'm the Exception. We're a 500 person company, the majority of whom are senior engineers, so people with around 15 years of experience. So now what we're focused on is AI, native engineering. How do you use AI throughout the software development life cycle to create production grade and secure products and platforms for enterprises. And so that has been a longer. Journeyform's always been known for kind of adapting the latest technology. This is kind of the latest chapter of that. But there's a lot we're doing internally. I can talk about later on how that upscaling is going.
- [06:41]
A
So at the top I mentioned data. Right. There's also a new pitchbook report. All of it together paints a pretty strong picture for Ireland. But this didn't happen overnight. Right. So I'm curious, from your perspectives, what are one, maybe two factors that you think shaped the country's success in the space?
- [06:59]
B
Overall, I think there's a few things I remember back in the day and I'm not going to date myself. When I was in grad school, we learned a lot about the Irish model and the role that the government played in certain tax breaks and so forth to incentivize the world's biggest tech companies, many of whom were America based, to come establish their European headquarters in Ireland. Now you know, that has had its ebbs and flows over the years, but I think that infrastructure of like, wow, the world's biggest tech companies are there. And I know myself, I'm, I'm Turkish American. A lot of my friends from Turkey have moved to Ireland because there's great engineering jobs there. So there's a talent basis from kind of historic investments made that I think is a really strong foundation. I think the Irish government even now continues to be at the forefront of a lot of kind of adapting AI where relevant and kind of using the latest technologies. We, we helped Ireland build the COVID app. I think that's five, six years ago this week weirdly enough. And we did it using open source. It was quickly used by New York and New Jersey. But the Fact that the government of Ireland said, can you do this in two days, we don't have weeks. And we're very keen to move quickly and embrace the latest technology. Even then, I think indicates their approach now. So I'm pointing a lot more to public sector than private sector just because that's been the bulk of my experience. But I'm curious what you think.
- [08:26]
A
So as a longtime business reporter, I'm always trying to better understand that cultural nuance we just brought up.
- [08:32]
B
Right.
- [08:32]
A
Like what do people bring from their backgrounds? How does it shape the companies that they're in? And I'm curious from your perspective, like how you see the difference between working with American companies, working with Irish companies, how does that manifest if it does?
- [08:46]
C
So what I've actually my experience and it's, it's across maybe six or seven different states is there is a fantastic spirit of embracing the entrepreneur, encouraging the entrepreneur, helping in any way possible from the American people. I've been blown away by it.
- [09:10]
B
Yeah, I'd say from kind of the last couple of years of my experience. The it's, it is interesting. So nearform is we're about 500 people split across kind of Europe and the Americas. I do notice that the U.S. kind of, we work with regulated industries, enterprises in that space seem to be moving much more quickly in terms of AI adoption. I would say successfully necessarily, but much quicker to. Let's try pilots. We're already talking agentic, you know, with folks in the US whereas our European clients, especially in, you know, financial services sector and so forth, which are for obvious reasons not risk loving entities, are a little more cautious.
- [09:55]
A
All right, we should talk about AI agents. I mean, we are at south by Southwest actually. I think there's a rule this year, right. You can't be on a stage here unless you're talking. No, no. I mean. But do either of you in seriousness use agentic AI right now? And why or why not?
- [10:10]
B
What we're doing is called AI native engineering, where we're taking ostensibly spec driven development and basically developing agents that can execute a lot of the manual coding that heretofore was obviously done by people. And what that enables is massive, massive efficiency gains. And that's probably the first thing that clients talk to us about is like, how do I save cost, especially given all the, you know, economic uncertainty, how do I get more efficient? So what we do is spectrum and development. Why I'm a big proponent of it is it enables using agents in a very thoughtful way. I actually call them trusted agents, which might be my marketing spin on it. But there's no. As opposed to a black box, it's really a clear box because it's human in the loop. It's human overseeing the loop, where you're able to articulate kind of specific parameters, guidelines, let the code run, everything is spelled out. It makes it much easier to audit, to adapt and to use. So we use agents constantly. One of the approaches we really like is bmad, named after Brian Madison. We just got to talk to him last week. Very cool guy, by the way. And that ostensibly is kind of a series of agents where each one basically has like an Agile team Persona. So I'm an architect, I'm a scrum master, and each of those agents runs sequentially. So you're basically what we see is, like I said, our whole team is senior engineers. And so the role of the engineers, I think fundamentally is shifting to more of the way Kieran, our CEO, puts it, is the role of a conductor. You have to think really hard up front and you have to lay out those guidelines and really embed kind of some degree of governance throughout, which we can talk about later. But also, it's just changed the process of engineering. It's changed the software development life cycle. So I'm a big advocate of the thoughtful use of agents, I should say. I'm also an AI skeptic in a lot of ways, which, you know, I'm happy to share my thoughts on. So I think when done with the proper oversight and kind of proper data trading and so forth, there's a lot of opportunity here. And we're just getting started. I mean, I think, you know, a year from now we're going to be talking about swarms, not agents.
- [12:34]
C
It is moving very, very fast. And I think the way we use technology now is changing and, you know, it's being improved every single day. So you said that earlier. Like what you're looking at now is, you know, 24 hours time, it's taken another leap forward.
- [12:52]
A
So what's something that's been harder about this whole tech evolution than you really expected it would be?
- [12:57]
C
I'm not a technician, by the way. As I said earlier, I'm an accountant and I'm the one who's designed the whole product and designed the whole lot. So I know what I need to know. Technology, I guess. But I know when ChatGPT came out, my personal expectation was that that was the holy grail. That was it done, and we didn't kind of really need to go any further. And I was like, great, you know, we're there, right? And then, you know, kind of hired, you know, more on the team to actually build what, what we've now got kind of three years later. Right. And I guess for me it was, it was understanding the limitations of technology as to where it could do. You know, like, you go into ChatGPT now, for example, you can ask a question and it gives you great, you know, some great answers. But if you were to go in and ask a questions about, for example, your own business or your own, you know, your own data, it gives you general feedback. It doesn't give you specific. Even though you can upload reports and you can upload stuff, you'd have to upload everything going back to beginning of time to get those insights. So I guess for me it was, it was recognizing that that pivotal moment when that was released was only the start. And, and, and, and because I thought that was it. Great, we're there, you know, and, and, and that is, from my perspective, it's
- [14:22]
A
a really great point though. And it sort of illustrates this dichotomy of a time that we're living in where, you know, most of us drive cars without knowing how to rebuild the engine or a whole lot about what's under the hood. And, and at the same time, it's true that that expertise and that nuance, that's our distinguishing thing at this point, because the baseline general will be available to everybody with these tools. So how do you get there, how do you wade through is going to be part of the story of our time.
- [14:49]
B
I love that point. And I'd also add, you know, ChatGPT vibe coding, right? These are things that have made AI very accessible to everyone. I think it's actually, it's also brought with it a little bit of, not backlash, but kind of, it's made people think, think that AI is easier to implement than it actually is. So it's like, here's a tool, all right, I'm going to use this instead of Google or searching on Reddit from 2012 for a question. But the biggest challenge that I see both in any organization, you know, not just our, you know, enterprise clients, is you can't throw a tool in there and think that's fixed. It's people and its process have to fundamentally be, you know, people need to be upskilled, processes need to be redesigned, incentives might need to change and so forth. So it's this notion of a standalone tool that those kind of, kind of teased us with that actually are pretty antithetical to how you Actually need to think about integrating AI into a, into an organization.
- [15:49]
A
Oh, agreed, Totally. About this time last year, I felt like I kept saying over and over again, guys, no company cuts its way to greatness, right? Where's the efficiency is fine, but where's, where are the moonshots? Where are the dreams? And also, like, be real for a moment. Once we vibe code this thing, if it stops working, who's going to fix it? Who troubleshoots? Who understands it? If we don't bring up the next. You know, I think now we fast forward a year. The analogy I hear all the time is like, oh yeah, you could get a Model T, Ford and any color you want, as long as it's black. And then, right. This idea though, that we would somehow be settled or that now with these new things, we're just going to be settled to accept what exists instead of taking that and running forth like General Motors comes along and you can have any car in any color, right? Like, this idea that we're going to stop dreaming and building is sort of silly. It's just not where we are. What's something you've changed your mind about?
- [16:39]
B
I was very much an AI skeptic, which is a weird role to be when you are telling the story of an AI engineering company. My mom always says, I hate AI, but I love chat GPT, you know, And I was like, you can't have it both ways. So what I, I think about it and there's something around, look, films were using CGI before and now they're using AI, you know, to make those special effects even better. You know, where do I draw? Why am I drawing an artificial line between the approach to how something is done if it's actually better and the effect looks better and things like that. So if you translate that into kind of technology and development and our sense of engineering, I had a lot of the same concerns that I think the media loves to kind of clickbait you with, with agents are taking over AI this, hacking that, ah, the world is ending. But really, if you. And I'm not trying to disparage anyone, but like, I, I felt like I used to be more ignorant than I am now. I'm certainly not an expert, but I do understand that AI does not equal hacking. Death of an organization, all risk, no upside. I, like I said, I genuinely believe that a lot of the risks and the concerns that I had previously had have been assuaged over time. You know, kind of both in personal as well as professional deployment.
- [17:56]
C
So, so I set out Because I had spent so long in the accountancy practice giving guidance, given, you know, whatever consultancy to businesses. When I set up SOBE Analytics, I set it up saying, I am not doing this anymore. I'm not, I'm going to automate everything. And when, when AI comes in, it's all automated. You know, we've got accountants and business owners. You know what, you can just sign up and follow the yellow brick road and at the end you're going to have, you know, this amazing whatever experience with guidance. And so what I've learned over the last couple of years, particularly working with accountancy firms on the larger kind of accountancy firms who are starting to embrace client advisory services and you know, businesses who are demanding it. So there's a real market there. But, but what I learned was that the accountants, first of all, they didn't want to learn to be shown how to do it. They actually wanted it done for them. Okay, so then that kind of brought me back around that I said, right, let's look at this product again. This can't be a kind of, you know, a click through service or a click through. You just sign up and you get the results. Because the layers of the data within a business are complex. And understanding the relationships between one part of the business and another does a complexity. And that complexity, okay, we can put AI in on top to kind of sort out and analyze and trend and so on. But the understanding and the implementation and the execution of the guidance that's coming out of SOBE needs a human touch. And this actually was a bit of a, I guess an eye opener for me because I think when AI came out at the beginning we were all hearing, oh, all the jobs are going to be gone and AI can do everything. And it's this all seeing wonderful thing. And for me it was actually looking at it at the coal face and saying, no, it can do a lot and it can do a lot of an analysis and it can do, but it can only work with the data that is available to it at a particular moment in time. It doesn't understand what's actually going on maybe in the business owner's world. So if you were looking, for example, you're using AI and you're saying, okay, tell me about the health of my business and you've got a problem with cash flow, AI even in Asobi is going to say, hey, you've only got two months left to make payroll or whatever, you're running into trouble. It doesn't know that the business is actually about to release a fantastic new product that's going to bring a huge amount of money and a huge. Right. It doesn't know that stuff. So therefore, I guess when you say what I've learned or where I've changed my mind, what we've done in as Sobe now is I would say that we've got 80% AI, but the 20%, it's human intelligence. And that has got to be filled by people who have expertise and who actually are able to communicate with the business owners and communicate with users. I think that that's actually a curve that's kind of come around. So it's not that all the jobs are going to be gone. I think, you know, the expertise and people who are using their experience are going to be more in demand now than ever, in my opinion.
- [21:23]
A
So there is a floor mic over there. If anybody in the room wants to join this conversation, please feel free to walk on over to it. Why don't we talk a bit about what's coming next for each of you as you look towards the future here to the extent you can tell us.
- [21:37]
B
Well, I'm very excited to break some news from the near forum side of things. So we, you know, as I've mentioned, for 15 years have kind of been doing best practice senior engineering and we've realized that along with AI there's, it's, it's not business as usual anymore. Things have changed and we're getting a lot of frankly similar questions from so many different stage and sectors of companies. So today we're announcing that we're launching something called Nearform AI Factory. And so what this is, is kind of a culmination of our learnings from the last year and basically trying to accelerate time to value for enterprises. Basically we've made all the mistakes that you shouldn't and so let's get started to get you value. So basically there's three components. A value solution where we always start with a business challenge and then kind of take a thin slice approach from there. A solutions lab where we do what? Basically the step where most enterprises get stuck here is kind of proof of concept pilot, but avoid the POC graveyard through our third component which is a production office. How do we actually get things in people's hands? And I think that's what Nearform really prides itself on. We don't just do POCs, we want everything to prod, whether it's for a consumer or an employee or a developer. And all of that is built on kind of a foundation of strong data engineering. And then we have a kind of proprietary library of agents and components that we use. So very excited that we finally have an offering for enterprises that, wherever they are in their journey, and I'll be honest, between us girls in this room and no one else listening, no one has everything figured out. There's, there's kind of room to just accelerate and de risk and take some of the pain out of that AI adoption process for enterprises.
- [23:25]
C
So as you would have gleaned from what I've said up to now, our focus is very, very much on getting information into the hands of business owners to help them make decision making, you know, data driven decision making, quicker, more effectively and so on. And so what I actually see is, and it's not so much, you know, looking at technology support, but I think there's going to be a massive culture change. So I actually see business owners being able to look at their phone, using asobe and checking the health of their company. And not just, you know, now we can do it for the health. We can all check our own heartbeat and we can check so many different things. But right now you can't. There's nowhere to go to say what's going on in my business. What, what do I need to do today? And that is what we're driving towards in Ask sobe. That's what I want to see in the future. I want to see change.
- [24:18]
A
All right, we have a couple of questions from the floor.
- [24:22]
D
So number one, thank you all for being here. Secondly, in a way, you've already kind of answered my question, but I feel like since AI became not so much introduced, but from the moment it became popularized, everybody fell. If you didn't follow in love with it, you were scared to death of it. And we're, we're, we're miles away from sentience. We can all agree on that. But at the same time, what I've been seeing in the past year or two is the trend of approaching AI from a parenting perspective. Because you think about AI is this, is this, is this. You think of AI as this quote unquote thing that has access to all the information in the world. Yeah, but, but at the same time, like a child, as soon as that child becomes conscious of everything around it, you have to guide that child through the things that it discovers. And even now I was asking Google Gemini a question and it answered, well, first of all, I'm just a predictive model or what have you, so you have those safeguards in place. But that being said, how important do each of you Feel if that is the case, if this is the proper analogy, how important do you feel it is to stay on that path? And do you think I will eventually get to the point to where, like a child, we no longer have anything else to teach it and it knows to. And it moves beyond the place where it is now to where it doesn't exactly know how to apply nuances, it doesn't learn, it can't think laterally through experience like humans do. So, yeah, how do you, I mean, how do you think, where do you think we're going to go from there as far as that's concerned?
- [25:53]
A
Concerned Guardrails are incredibly important for a bunch of reasons. Right. Also, this is what makes AI, in a way this fraught topic. There is no definition for artificial intelligence, there's no definition for intelligence, and we have actually no idea what sentience is. So we model this off the brain, which we also don't understand, although it remains the most powerful computer on the planet.
- [26:19]
B
Yeah, I mean, it's such a good analogy. I hadn't heard that put in that nuance before and I really like it because if you take it further, you know, I have a two and a half year old niece and whatever, you know, I, if I say a bad word, she knows, she learns. I didn't even think she heard and she says it right. So AI has a little bit of that. Not to dumb it down, but whatever information you give it, it has a disproportionate impact on kind of its next steps and its outputs and so forth. So there's a big emphasis and we see this with kind of the intersection of open source and AI. A lot of the kind of space right now is on the weights. It's not on the training data, it's not on the inputs. And so there's thinking there that I just think is very nascent. I don't think we figured it out, but I do think the next natural step of creating an agent is creating, you know, an agent to create that agent. Right. There's this will continue in some degree of weird meta. So I think there's something around, it will continue to get more complex. And I, I do think we need to be wary and have guardrails.
- [27:20]
A
Well, I think that to go back to my car analogy, inventing cars helped us figure out that seat belts were a good addition and we're going to have a lot of learnings like that too. So that's just the nature of being, you know, human and alive.
- [27:33]
B
Hello, ladies. Just first off, wonderful to See women in stem on a panel. So glad I wandered in. Lacy Rapini, I am a technologist with Worldwide Technology. Congratulations on your big releases. Curious to hear a little bit more about this clear box you talked about with agents. So I know recently, you know, I was watching a YouTuber, he built an agent. The agent wanted to be better. The agent took his credit card and enrolled the agent in an agent class. And I was just like, first of all, but like, tell me a little bit about these parameters that you're building for this clear box, thinking without leaking secrets. I'm just very interested in it. Yeah, I mean, there's been a lot of scary news lately and agents that are overreaching, should we say. But again, it does come back to guardrails, which I'm not trying to say that to be flippant, but the way I think of it is not just a tech stack, but a trust stack and having that built across the sdlc. So what does that mean? It means having mechanisms for observability, measurability and so forth at every step of the way. The reason the agent is doing that is because it wasn't told what it can and cannot do correctly upfront. And so what is either the training data and, or, you know, the instructions that you're setting up that agent with. So I think, you know, to your point, we're going to, we're going to see mistakes made and kind of learn from those, which is not a great way for me to suggest doing things. But I think the more things are documented, I mean, previously things just lived in people's heads and then they would type out 10011. This, I think, is much like, you know, spec driven development. You know, AI native engineering is much improved to that because all of a sudden there's documentation and, you know, you don't need to redo a whole bunch of source code. Instead you just kind of have this natural language document you can work on and so forth. So that's why I see it as a clear box. If you set it up in that way with a trust stack emphasis, not just the tech stack emphasis, I guess.
- [29:33]
A
So we need to wrap up because we're pretty much at time. But I have one last question for you, and that is, from where you sit, we talk. It's one thing to talk about all the data and all the trends and things like that, but what's one thing is make one trend, one decision, One factor that you think will help determine whether Ireland remains one of Europe's most concentrated Innovation hubs, let's say 10 years out.
- [29:56]
B
I think the public sector is doing a lot in the private sector there will always be. As far as AI adoption, you kind of see it in two directions, right. You see grassroots developers are naturally curious, amazing people. They're already using AI, whether legal says they can use ChatGPT or not.
- [30:13]
A
Right.
- [30:14]
B
It's the leadership buy in, the C X O level. Not just CIO but also CEO level of setting AI and whatever that means, AI literacy upskilling and adoption as a priority. And not just we need to use AI, but really rethinking their operating model and business model. There's a great book on this called Competing in the Age of AI that a former professor of mine wrote. And it's really rethinking the entire value creation, value capture, the entire kind of chain in that industry. Right. So I think if we can get it, I don't know how to do this, but buy in from the leadership side for that top down encouragement and also governance of AI in some of these organizations. I think that's the missing, well, not the missing piece, but the opportunity right now.
- [31:04]
C
And I guess I'd kind of circle back around to what we were discussing, you were talking about earlier on and we were saying about, you know, the creativity and the amazing ideas that are actually coming, you know, from all around the world. But like we're talking about Ireland specifically. A lot of these young guys and gals are coming out of, they're spinning out of universities and some of them aren't actually even finishing their education because some of their ideas are so mind blowing and you know, it's giving them the support and encouragement which I know enterprise Ireland, you know, are amazing at. But then I guess if we are focusing on Ireland again, I'd swing back around again to say about the funding side of it, you know, which there, you know, there is a big hole in funding in Ireland. It's very, very hard to get the funding. And I would also, and I'm sorry to play defense feminist card, but also if you're a woman looking for funding, if you've got females and a lot of fantastic technical, female technical gals coming out and they are not getting the same chances, the same opportunities as their male counterparts. So when you kind of say about what does the future and what does it look like, I think that's an awful lot of those grassroots issues that need to be addressed. We can kind of talk and dream about AI and all the wonderful things, but we've got to help people who've got those ideas to just give them a chance to explore them.
- [32:38]
A
This has been a fascinating conversation. Thank you ladies so much. I have learned a lot. Also, thank you to the Room for those delightful questions. You've given me a lot to marinate on. Thanks for listening to the next innovation. This series was produced by Situation Room Studios and powered by Enterprise Ireland. Investing in the next wave of innovation. Our executive producer is Christine Barata and our senior producer is Sharon Barero. Additional production assistance by a global situation group. I'm your host, Jennifer Strong. Until next time,