AI’s next test: turning pilots into productivity

0:00Freshback from a journey into the belly of the AI beast in San Francisco. This is the business of tech and IPdegriphant. Yeah, I've just returned from Dreamforce, the annual conference of the software company sales force, which drew around 45,000 people to the Moscone Center last week in San Francisco. It's always been a barometer for where software and enterprise technology is heading. But this year, it felt like something more, a gathering at a decisive moment for artificial intelligence. Onstage were some of the most important figures shaping the artificial intelligence era. Nvidia's Jensen Huang, the chip master, open AI's Sam Altman, andthropic chief executive Dario Amadei.

1:01And Salesforce founder Mark Benioff presiding over it all. He's the one who's gone all in on AI with agent force within the Salesforce platform. And last week, touted a major partnership with an anthropic called Claude Force, which allows the Claude AI assistant to become the gateway to everything in the Salesforce stack. The message was pretty clear from that, AI will become the go-to way to use software suites that you use to need to learn inside out. But amid the excitement overall of this came a warning. Amadei and others have argued for greater industry coordination, oversights, and caution as these systems become more capable. At the same time, businesses are being told that they've barely begun to tap the value of AI tools already available to them. Dario Amadei, for instance, who penned that memo a couple of weeks back calling for a slowdown in the pace of development of frontier AI models, suggested we're only utilizing 5% to

2:0610% of the capabilities of the existing models. So he wouldn't really miss out on much if the industry collectively decides to dial back the breakneck pace of change. President Trump has very different ideas on that. He's opposed to a slowdown, but he is setting up an AI force, like the Space Force, and will appoint an AI SAR to oversee the government's efforts to guide the industry along. Whether they'll just tell him what he wants to hear or challenge the president to take the warnings of now numerous hundreds, thousands of AI scientists seriously is yet to be seen. So are we moving too fast or not fast enough? And what does this really mean for New Zealand businesses anyway? Were companies such as Zero, Fisher and Parkle Appliensis, Halter, One in Zed, and others are beginning to put AI agents to work in customer service, field operations, knowledge management,

3:08and really innovative ways. My guest is Richard Filmore Smith, Salesforce New Zealand's country manager. He's been in the role since taking over from Hamish Miles in February, but he's been with the company since 2020 following around 18 years spent at a number of key roles at Vodafone UK and then at Vodafone New Zealand. We sat down at Dreamforce to talk about the rise of agentic AI, the future of software interfaces, why trust were the data really matters. And how businesses can use AI to lift productivity without losing the human connection. So here's my chat with Salesforce's Richard Filmore Smith right here on the business of tech coming up just after the break. Richard, welcome to the business of tech. Thank you, thank you, Peter. Appreciate you being over here in San Francisco with us for Dreamforce and looking for this conversation. Yeah, I think it's my fourth or fifth Dreamforce and it seems to this one in particular at a

4:10really pivotal time and played out in that first days series of key notes. We had all the leaders of AI here. We had Jensen Huang from NVIDIA. We had Dario Amade from Anthropic, big partnership announced this week with Salesforce, of course Sam Altman as well. Mark Benioff himself, who has been one of the most aggressive moving into AI as a software leader. Interested in your take on the zeitgeist of the vibe you're seeing, this sort of discussion around how fast we are moving with AI and obviously Dario's memo came out just a few days before Dreamforce kicked off, calling for a slowdown or at least collaboration across the AI players to coordinate their efforts around safety and oversight. On the other hand, you've got Mark saying the door is open for customers to use AI but not all of them are going through it.

5:12And you've also got Dario saying we're only using 5% to 10% off the existing capability. So some conflicting messages there, what's your take on where we are at now in this journey of artificial intelligence? I think my son is, wow, what a week. Dreamforce is always phenomenally exciting time and I think this week for me like you, I think this is my fifth, is probably the best ever, the energy out there is palpable. I think when I look at what's also fantastic is the representation of our theorem, it has been fantastic, representation here from our customers and partners and that's awesome great to see. And I think, for me, I think it's an interesting, that's a matter of exciting in terms of the next phase of this journey of a Gen.T.K.I. And we talk about some of those conflicting pieces there. I think the most important thing here is that there are probably two big things I'll talk about. One is the fact that we are moving to this more open system of allowing what we call AI

6:15Force, which is going to open up and say, look, Salesforce has continued to deliver phenomenally strong platforms, the power of Salesforce is in that data, it's in the workflows, it's in that business logic, what we call the deterministic layer. What does that give? Deterministic means it gives organizations confidence that they'll be a repeatable outcome through that deterministic layer of Salesforce platforms. With AI Force, what we're now saying is you can create this live interface layer and you can bring to that the interface that's going to relevant to your business and whether that be Slack, whether that be Claude, or that be co-worker. I think what that's giving is in the time when things are evolving, we're giving optionality to our customers. That to me is exciting. We're saying the customers build on the trust of Salesforce and take that optionality there. I think that trust piece is the next bit. We talk about the next phase and this opportunity with AI. Organizations need to trust and know that what they are providing to their employees and their customers is information that is relevant and is grounded in their organization's

7:20data. I think you've heard Mark say it. Look at the camera during the keynote and said, your data is your data. We don't use it to learn. We don't let anyone have access to it. You have the control on the guard. Zero retention, they call. Exactly that. I think that's one of the things that when I talk to customers, they absolutely latch on to. Kiwi businesses want to know that they have confidence in Salesforce platforms up. They don't have a lot of founding values as trust and that trust piece runs right the way through the AI technologies that we are making available with our Azure Force. It's interesting because there's been several studies in New Zealand. One in T.A.I. Trust reports every year and they find it trusts in AI as low. When they ask, average Kiwi is not business people about it, but Kiwi's wise your trust low. It's their user experience of AI to date has not been good. They've used a chatbot and it hasn't worked or something like that.

8:23It's not necessarily about the data center. It's the bigger picture issues. It's just a frustration. What do we need to do to actually improve that user experience? I've seen like CVS, the big pharmacy player in the US doing amazing stuff with the automating prescriptions and doing all of that through a chatbot, logging in with your details so it verifies your identity, knows what scripts are coming from your GP or that sort of thing. Is that going to be the point you think where Kiwis go? Actually I do trust this. This is genuinely useful to my life. Yeah, and I think that starts with the quality of the data. That's the reality. It sells for us. We have data 360, data cloud. That is fundamentally, you look at those that have been successful. We started off with a little bit of that Fishing Piper appliances and what they have started with is basically they've got their data cloud, data 360 is grounded with thousands of Fishing Piper existing knowledge articles that provide product insights, guidance on how

9:27to use your appliances. What AI is able to do is to surface that to the consumer in the way that they want to be met with the most relevance and that's allowing them to, I think they've seen, they can now do almost 45% of their service bookings can be completed online. They've seen 50% reduction in call handling times because the information that's available via data 360 and servicing that information to the customer is happening in a much better way, much more coherent. So it has to start with data. Businesses, all of you get the data right. AI as a tool and then to surface that in the way the customers need it to be surfaced. Yeah and obviously anyone who's bought a dishwasher or a fridge can use that interface. I was really intrigued this time by the Technician Workforce tools, AI tools. So when a technician goes out, the AI is suggesting, okay, this is the fault you're going to fix, you will probably need all of these parts as well as to telling them where to go and pulling all of that information through for themselves for so far.

10:31So you need sort of service industry player where you've got a field force of people. It's increasingly valuable too. Absolutely. I think what you're talking about, I love about that story there Peter, is that's about the human and the loop story. AI, we know that we have got an opportunity to lift productivity. We know we've got an opportunity to lift customer experience. We know we've got an opportunity to lift employee experience. That AI scenario there is fantastic. You are a service agent, you're out on the road, you're a technician, how do we make sure that all those workforce people, they have that right information at the right time, it gives them a better experience and then what they can deliver to the customer, the customer of Steven Service gets a better. That to me is the perfect example of where a GenDK AI comes together with human and the loop. What struck me this year, and it was actually a comment from one of the Aussie analysts series, when we were in Australian New Zealand sort of press conference, he was like, why aren't you showing us the Aussie customers, it was obviously zero and Fisher and Plycle. There are Aussie customers here, but these are exemplars and you've got halter here,

11:36health New Zealand is here as well. What is it about, it seems to be when it comes to Salesforce and use of particularly agent force and AI implementation, why are New Zealand companies, which we're not considered like a powerhouse of AI, but they seem to be willing to adopt it, at least some of these companies, one NZ being another one and doing really innovative things with it. Yeah, look, I think we, as we do, frowly in LTORO, we've got QBizzi Lee in the way and this is something that personally and collectively, there's great pride in that. I think we have seen that play out, you say, zero and Fisher and Plycle appliances have absolutely been up there on the big stage, you can see 40-foot head shops of some of our customers up there, which is fantastic. It's become from many years of partnership with businesses and organisations in New Zealand with Salesforce has enabled them to deliver great customer experiences on the Salesforce platform, so that be sales cloud, service cloud, market cloud and others.

12:36What that's done is because they built out very strong platforms with the right data constructs, they've always thought in a very customer-focused organisations. Agentec AI and agent force is really the evolution of that. How does that now move to being a way we just serve our customers in a different way? We enable our employees in a different way. That's Coffee Supreme, for example, who use sales agents to ensure that every time one of the Coffee Supreme reps shows up, they've already got all the insights about that business they're going to see all served up by agent force. I think we're seeing organisations who've got, we've thought well about their strategy around how they want to use data and how they want to serve customers. I think QB businesses have absolutely lent in to piloting early. There's not a single customer out there we have relationships with who aren't saying how do I get piloting? Now what we're seeing and what's great is we see more of those customers move from the pilot into the scale. That's I think where we're sort of not too big, not too small, size of QB businesses have allowed them to take a very broad vision, get excited and

13:38really get out there. As you say, we've got some great stories that are being shared around the world, and that's something all of us in New Zealand should be really proud about. Yeah, there are these sort of spotlight companies that are leading the way. And I think we can understand how important that is to inspire other New Zealand companies because we're probably on the scale of things as a nation, we're not the biggest adopters of AI yet. There's a lot of piloting going on, as you said, but it tends to fall down a little bit, spreading that across the business. I've been to a couple of sessions of some of your colleagues here talking about what is required to sort of, you know, bridge that gap. And it really seems to me comes down to building that capability in your organisation. And one of your colleagues gave a session yesterday, the five new roles that every organisation will need and they're all around AI and three of them were about skills, workforce, developing

14:40your workforce, as AI operations and things like that. But is that something that we quite haven't focused on enough in New Zealand, building that capability within our teams, upskilling people, giving them these skills to actually confidently implement AI? Yeah, look, it's up to the user journey. And I think we're in a very exciting time where we're seeing a lot of people coming into the workforce who are AI natives. I think that's the real opportunity to look. We know that theory is a great at adopting new skills. And I think that we look at the opportunity there is to really tap into that and get those coming into the workforce who already know how to use these tools and get them in organisations and getting them to sort of supercharge them. I also think we've got a fantastic collection of partners in across New Zealand who have this skills. We have sales force, our sales have evolved our model and we're moving to this model of what we call FTEs for deployed engineers. They are people who are actually now getting into a customer's business and helping transition

15:42from, we've got a vision, we're building out a pilot. How do we demonstrate that pilot can absolutely work in the way that we expect to do this grounded in the data, it's servicing up the agent force autonomous AI experience in the best possible way. And then it gets really good. Now we've shown that, now let it skill people up in your business or with partners to then take that right the way through to scale. So I think we've absolutely got the intent from customers to skill up and taking a broad view of how to skill people. We've got partners in country who do a phenomenal job and have built zero's example of their zero essential assistant agent was a piece between zero and a partner. And that's really showing that the power of that. So I think all in all I think we're in good shape. But yeah, and I think we're at that kind of the tipping point now. I think with the confidence and is absolutely their businesses organizations can see that the technology does work. It's trusted. And I think we're now seeing an upskilling coming off the back of that.

16:50You talked earlier about AI force, which really is the big news of the weak. The board force had been launched a few weeks previously, but talk us through the shift that this means it's been a lot of talk about what is the future of the software industry. Your colleague Patrick Stokes stood up in the keynote on Tuesday and basically showed an interface that runs his part of what he's responsible for in Salesforce, a dashboard of everything that is important to his job. And that looked very different to any Salesforce interface I've ever seen. So Mark Benioff was basically saying you'll take Salesforce to wherever you want it to be. It could be on Slack, which is owned by Salesforce. It could be in Salesforce, but it could be in the Claude interface and increasingly probably open AI and others as well as part of AI force talk us through what's involved in this fundamental shift and what's driving it. And I should also say to Ress or anyone out there who got Salesforce date also could be

17:51existing Salesforce UI. I think the point is the optionality there. So I think the exciting part here is where AI and SAS comes together and software comes together is in this AI provides its probabilistic capability. It can make decisions based on the insights in front of it. But we all know that businesses operate with a need to have a consistency, a deterministic outcome. I think this is the absolute exciting part about the AI force announcement is that we are saying Salesforce is saying that our platforms continue to provide everything they have done for 25 years around the trust, the security, the governance, the semantics, the need, all those workflows that your business runs on are still there. But we're also recognizing that that front door to come into that, each business might have a different need to how they want to do that for their employees and their customers.

18:52So in its simplistic terms, AI force is saying that this that live interface layer could be slack, could be Claude, could be co-worker, could continue to be existing UI that Salesforce has. So again, that's the most exciting thing and I talk to especially our CIOs, CTOs who are here like this is a game changer. And Salesforce, as I think Patrick said, we have led the way from a software organization to say this is a complete and utter open way of accessing. But it's come from years of architecting our solutions in a way that gives the ability to maintain the trust, maintain our security and have access in. And I think that's the important part is then we start to talk about the AI harness. That's about extending the trusted capability of Salesforce into those interfaces and showing that your data can be controlled in the safest possible way. Yeah, one of the buzzwords of the week, it's not a new concept but headless architecture.

19:57I tried to steer away from that. Yes, you are right. The part of the history or the definition of what that means in Salesforce context. I mean, headless is really in that simplest way of saying that the underlying system is in that system of record continues to be there. But the headless mean means it's open and what kind of access that and how you come in. I mean, headless is the technology term and we've obviously coined the AI force has been the overarching piece for that. But that's simply what headless is. It doesn't have to have the traditional UI at the front door to that platform. It's incredible, like Patrick said, he spun up that interface in Claude in six, eight minutes, a couple of days of iterating, getting all the data feeds right. But we are going to potentially be in the next few years in a situation where everyone using Salesforce and other platforms as well is going to have their own tailored user interface.

20:57We won't all have to sit down and learn a system. What does that mean for the tens of thousands of people around the world who are trailblazers, the name for your super users of Salesforce? It's quite a shift. It's probably liberating, but it is. People who have known the system for 20 years know that ins and outs of it are Salesforce administrators. What does it mean for them? I think we heard the keynote. I think there's probably most exciting thing for them because all of the work they've done and everything they have done to train and understand the Salesforce architecture and the platforms that they're continues. Really we're just talking at the front door. That's the piece. This is a big shift we've seen in terms of the understanding of how a genty AI comes together with software. Is that actually everything we need in that deterministic layer, which is you talk about those trailblazers, they are in their evangelizing how Salesforce can supercharge their

21:59company and they will continue to do that. What we're now talking about, and we know many of people have used Claude or chat GPT or Slack know how simple it is to use. That's the exciting thing is we're saying that there are going to be also multiple activities you're going to do at the front door and then they will pull on the Salesforce, they will call on the Salesforce system to present that information to the employee or the customer. I think admins are going to make an admins or trailblazers are going to become more important than ever to really understand how those two things come together. One of the other terms I've heard this week, and actually I was at a session with zero and Crocs and Pearson's, big clients for Salesforce, the idea of super agents and Mitch from zero is basically saying the past has been a monolithic agent. One big one trying to do everything or you have sort of siloed agents. What is the concept of a super agent? I think we saw the number of activities that that agent can do.

23:04They can, to that point, I think where we've seen the first runs on the board, if you like, with a gentigai has been to take a strong data set with a strong process and have then turn that into a way that can be identified and a gentigai can run across that. We have seen that in the number of the early use cases that have been shared. The super agent is done so okay. We're now going to have a single point of access that is going to perform multiple tasks, pulling on multiple platforms and multiple data sets that can then present back to the user a very clear outcome. With that being a scenario of, I'm changing by bank and as part of changing my bank, I need to move a whole load of things across. You can start to think about how a super agent might be to sit there and say, okay, we need to pull this information on here and there and we need your direct debits over there. You've got a credit card here. You can start to present that information and move as you move into a different banking

24:06that starts to prove that information across. That's a hypothetical example of where an agent there would have to go to multiple systems multiple data sets and pull that across to give a seamless, in that case, an aura of customer experience. Right. So do you think that's going to become the norm in organizations that are deploying agents that the super agent way is going to be a very efficient way to do it? We support each of our customers and what they're sort of vision is to go on this journey and each business will have different use cases, different opportunities, different risk profiles and where they want to start and where they want to get to. So I don't think there's one size fits all by no stretch imagination is there. Many, many are getting great successes by taking processes that will be enhanced through a GNDKI, with that be a process, an internal process to provide better information to their employees, with that be a customer facing process like we have with zero central assistant where that sits across a hollow of their content that has been traditionally accessed

25:08by by accountants or in customers that now has an agent running across that that provides you know, they've ended up with 70 almost 70% deflection rate, a C set of almost 60% by having an agent in front of existing sort of data that was operating in a sort of static way into now dynamic way. There's that right the way through to the scenario there are super agents and I think that will be part of the journey that each business will go on and we're doing supporting them on and seeing each of them sort of find their opportunities in the. Yeah, if you look at the New Zealand market interested in a snapshot of where we are where you see the growth and the opportunities, for instance, you know, fascinating stuff from the head of Siemens who's been at the shows done some really interesting discussions about manufacturing and the use of AI, you know, that is an industry that needs a bit of help in New Zealand. It's a tough environment for growth there and productivity is low. It seems like there's prime opportunity interested in you use on that.

26:08Any other sectors and particularly you're seeing a lot of opportunity and growth in the appetite for AI at the moment. Yeah, I think the beauty of AI is it's relevant across every sector. You touched on productivity there. We know as Kiwi's that where we sit and ask of the productivity stats and OECD level isn't not great. So there's a huge opportunity I think and personally that AI is going to actually just super charge our existing human workforce and provide opportunities for us to deliver better experiences. From an industry perspective, every single one of the industries we see is leaning in on it. I mean, particularly if you think about it, you've got a high number of customer facing people. We have, you know, sales sales agent there that can sort of provide bridge information then we talked earlier about the field services. You know, people like they feel service. We can service agent can enable those. The flip side is if you are have a high sort of customer touch. Now, with that being via commerce via via today, a service center, there's a huge opportunity to surface information to those people. And I touched on sort of zero there just but I think that's a phenomenally good use case

27:11for in terms of any business out there, any organization with a government commercial enterprise. You know, when you look at business, you can see use cases that are highly relevant, regardless of what your purpose is. If you've got a high volume of customer interactions, having an agent across that kind of that data repository, you know, that knowledge place is a really great way to provide a much better customer experience and take a load off your employees who can then be focused into, you know, other tasks, the high volume was in a scenario with zero. They handle handles of the straightforward questions and come through the very central assistant handles those where it can't provide the answer. It provides passes over to specialists with full transcripts. Just a really simple example of how a GenTK AI is improving both the customer and the employer. It's incredible. You know, that's probably, you know, that's available to five million customers. It's probably the, you know, the biggest implementation for a New Zealand company off of a GenTK AI.

28:11So it's incredible. Yeah. You talked about government there and I know that's not really your area of responsibility, but, you know, that will be probably the biggest area of transformation in the next couple of years. The government has signaled it once AI to transform governments and there's a mandate there to reduce the headcount. But what I hear is sort of talking to people around the show this week is it's not really a replacement for humans. It's supplementing, augmenting what they do, but redeploying those people to help implement AI. So what are your observations for any large organization that does have a lot of customer facing roles that is wanting to get efficiencies and productivity gains, but doesn't want to undermine or destroy its services by reducing trust or the usefulness of its services? Yeah. With your commercial enterprise or a government agency with a high customer base, everyone's leaning into the same objective are providing a better experience for the customer and

29:16providing better experience for their employees. And what a genty AI now does is give great choice of how you surface information up, how you surface workflows up to customers. If you can get that right, then what you can do is you, you can enhance the customer experience. You're also to that point about the productivity one, every business is looking at how to drive better productivity and to provide humans with the high value to provide that more humanistic engagement. We all type of need to go to a human thing of it with freeing up people to do that where they can add greater value to the customer outcome, that's an exciting place to be. And I think that's across all parts of every business and every kind of vertical that opportunity sits there now. Yeah. Well, I think that is the opportunity and Salesforce famously helped out the New Zealand government with the COVID screening app and database and all that sort of things. So there's some of the government stuff I've seen, the Texas, it's digital driver's license which runs on Salesforce is some incredible innovation going on in the public sector with

30:21AI. There is. And I think Mark put it up there. I think 50 states, all 50 states in the US have running some of the Salesforce to enable that sort of government outcome. So we're actually seeing across the board strong adoption in government, the same as we are in our enterprise sector. Have you got any view on where we're at in terms of our regulation in New Zealand? Have we got the settings right? Are we doing enough to promote businesses using AI? It seems to me like there's a lot more, even in Australia and that, a lot more promotion. They've got the national AI center. I mean, the prime minister has said he wants us to be really aggressive on use of AI. Does the government have a bigger role to play in actually promoting there? Look, I think that we're in support of having the right regulation. I think importantly, we continue to lean in in the times of the way in which our systems are showing up that provide the opportunity for customers to deploy a GND AI in a trusted

31:24way. They can trust it, the customers can trust it. And I think that that, that to us is the most important thing is that as a business providing these services, can you trust what you are showing and surfacing? And I think that's the bit for us is absolutely ensuring that we align to anything in terms of that regard. We feel we're leading the way in terms of how we're using our trust layer to provide that. The AI harnesses we call it to ensure that they're a unified, unifying approach to how you bring together the Salesforce platforms and the AI force together to ensure that people can be confident in what they're getting. That's the other buzzword of the week, harnesses, which have been around forever as well, but the headless harnesses and super agents. So lots of cool, cool terms to throw around. But just finally, as you head home, like I will be tonight or tomorrow on the plane, what's the key takeaway for you or something you brought away from Dreamforce this year that's going to change how you look at the New Zealand business? Well, I think I'm a firm believer anyway.

32:24I think in this saying, which is a rising tide, lift all ships. I think that's what I take away from here is that we've brought a bunch of our, our theatre customers, our partners together. We've given them access over this week to thought leaders and whether that be some of the leaders or some of the largest AI companies out there, whether it be with your Latina Robbins as well, you know, really thought provoking stuff. But also most importantly, what I take away from here is Dreamforce is about the human to human connection. The question I ask every single person I meet is, who have you met here that you couldn't have met when you're in El Quiro and how do you then maintain that, whether that be a Salesforce HQ leader, whether that be someone from our customers that are based in other regions, because the opportunity now is for us all to continue to learn together, share, develop that kind of these, these, these wins that we can get. And that's the bit for me. How have you gone away from Dreamforce was learning something new? How have you got new connections and how are you going to use them to supercharge the change you want to make within your business? And so I will be looking forward to getting back and do those conversations with the

33:27our customers back in New Zealand. Well, it's been a great show. I've learned a lot yet again here, exhausting, but my mind is crammed with lots of ideas. So thanks for having me and thanks so much for being on the business with Tech. It's a pleasure. Thank you Peter. As you can tell from my voice, it has been an exhausting, nice thing to go into finally on our show play. I appreciate you giving up the time to have this conversation, especially for being over here. I know people give up their time away from family and work to come here. So thank you very much. Well, thank you. It was well worth it. Yeah, brilliant. That was Richard filamore Smith, Salesforce and New Zealand Country Managers, speaking with me at Dreamforce in San Francisco last week. The takeaway from this year's conference is that the next phase of AI will be less about the chatbots and exactly what they can do. And more about how organizations connect their trusted data, which is at the height of

34:29everything they do, established workflows and skilled people to get useful work done. The technology is definitely moving fast, but adoption, trust and capability building will determine who gets real value from it. That was very clear from all the sessions that I went to at Dreamforce. That's it for another episode. I'm Peter Griffin. I'll catch you next Thursday, right here on iHeartRadio, Apple Spotify or wherever you get your podcasts.