Speaker 1There are a lot of companies moving quickly on AI adoption, launching pilots, training teams, and putting agents into work, which is great. But the harder question is whether the organization is changing enough around it so the technology's activity turns into real business value. So what we need to think about as well is, who owns the agent? What should leaders measure when agents start carrying out real work? How does the workflow actually need to change? And what happens when something breaks or the agent goes rogue? Welcome to Apply to AI Australia. This show translates AI complexity into growth, margins, and time. Today, I'm speaking with Leandro Perez, SVP and Chief Marketing Officer, Salesforce ANZ. Now, today's episode is going to talk about the gap between AI activity and AI value. Why leaders are moving so fast on the technology, but forgetting about the organization. Let's get into it. Leandro, thanks for coming on Apply to AI, mate. Glad to be got here. How's it going? It's doing great. Thank you for the opportunity. No, it's good, mate. You've had a busy couple of weeks, hey? You seem to be doing everything and everywhere. What's going on?
Speaker 2Look, I was traveling. I've been in Los Angeles, Chicago. We had our largest event for marketers in the year there called Connections. And then last week, we had AgentForce World Tour Melbourne, which is our second largest event, only just. It's behind Sydney. Over 4,000 people there. You know, there's a lot to communicate at the moment. It's an exciting time in our industry.
Speaker 1Yeah, 100%. I want to start with the tension a lot of leaders are facing right now. I mean, AI adoption is moving so quickly and moving quickly and at a high rate of speed doesn't necessarily equal value. So what I wanted to really unpack was what needs to change inside an organization before agentic AI or agents start to create? Real business value. So from what you're seeing in market, where are companies mistaking AI activity with AI progress?
Speaker 2And you're right. Many organizations are deploying, you know, Australia is leading the world on AI adoption, which is awesome. I think we, we ask service leaders about 76% of them are looking or using AI, but the real challenge to your point is getting it in production, getting value out of it. And our customers are moving past that pilot era and that's because it's grounded on what we've been building with them for many, many years, which is the context, right, and the guardrails of how to put something into market. And so, you know, that's where we're seeing true value. I think to your point of just spending money on technology is not necessarily answer. In fact, a lot of companies are spending increasing amounts on AI, but not getting that value. And it's because they're not necessarily grounding it in what the business already is trying to deliver. So maybe we can unpack some of those examples for you.
Speaker 1Yeah, yeah, for sure. And you're right. That was a bit of a loaded question. So obviously, you know, Salesforce has been around for quite a while, obviously large context with the customer and the organization, but if you look at it broadly from an industry perspective, MIT says 95 ish percent of AI pilots fail. You've obviously mentioned a few points, you know, multiple points actually, but from a macro perspective, not even just necessarily Salesforce, why do you think it is happening? What do you think are the top three reasons? Or a few reasons that come to mind? Because we're investing a lot of money in this and obviously clients want returns and some do, and some, you know, making repetitive mistakes that I often see as well. So what do you think?
Speaker 2Yeah. So firstly, there's an element of experimentation that has to happen with any new technology, right? So that's number one, right? I think to assume that everything that we tested was going to succeed, I would say, well, then you're probably not trying hard enough, right? If everything works, right? So there's an element. There's an element of that first, which is when new technology comes, you have to experiment because yeah, you have to work out what are the boundaries of what it can and can't do. But I think that was kind of the conversation of last year. This year with our customers, what, why we're seeing a lot of, you know, other competitive solutions not succeed is, and I'll give you an example, like I mentioned before, you put an agent on your website. If it's not grounded in the information that you have around what you offer, let's take an example, you know, like Fisher and Paykel, which I believe you met. Last week, right? They've got premium products, they sell appliances. If someone goes on their website and doesn't, the agent doesn't know the products they have, or even if it knows what they have, but whether they're available, or if there's an issue with one of those products, what are the recommended solutions to fix it? Well, then the person that's using it's not going to get any value. And then the first thing they're going to do is like escalate to a, to a human, right? Which is what we all used to do on the phones when we tried to get through those really complicated phone systems that everyone's still using, right? And so that's why we're. Very excited when we see that our technology is, um, for the many years and, you know, we, we're not a startup anymore. We started in 1999 is actually powering this agentic revolution. So let's continue on with another example there, right? High pages. They want to bring on a new tradie to, to the platform right now, to do that, you have to verify a whole bunch of their verifications. Are they licensed? Um, you know, what have they done that, uh, you know, for their record is, are they going to be a safe person in your house if you ask for high pages? So they're going to be able to, you know, you can't just bring someone around to your house, which is, you know, a very important thing for you. That system needs to be able to check against that. And that's what we've been working with high pages with the CRM we already built, right? What are their restrictions? What are the licenses they check? What's the information so that it can shorten the time. And that's where the real last mile is, is how do you get ROI and ROI comes from delivering this to solutions in areas that you really care about, right? So if you care about, you know, how quickly you can answer a service call, how much pipeline you can generate, which I can share some of that with you. Some of our own examples for what we've done at Salesforce, um, how quickly or how effective your marketing campaign is. That's how you then taro ROI, which is no different to how you tie ROI to an existing business case for any technology.
Speaker 1Hmm. So what do you think of the telltale signs that going back to, you know, the point before that organizations is just trying to look busy with AI and they're not necessarily ready to actually deploy it within their organization.
Speaker 2Yeah. So we saw a few signals in the beginning, which was a lot of CEOs. Getting pressure from boards to move faster. And you know, that's when you don't engage your team of just do X, right? It was like, I can imagine back in the day, which is let's just get iPhones. It's like, why are we getting iPhones? I don't know. They're cool. Let's get them. Right. And they are cool. Everyone had their blackberries at the time, you know, and they were very happy with it. Right. Yeah. Yeah. Maybe I'm dating myself there, but I think, uh, top down is good and you need that, right? You need that vision, but you need leadership within the organization. And this is where we're seeing the departmental. Owners, you know, the head of customer service or the head of sales or head of marketing in my case that understands the job to be done. And then how do you map the capability? Most likely working in partnership with a CIO or a technology department as well. All right. So I think you need both of those. I think secondly, what is also happening is that people are confusing what they see in their personal life with what they do in business. And that, by the way, we've been talking about this for years. This always happens with every new wave, right? You know, they want to search and then they want to bring search the enterprise and they realize, well, just search documents. It's very different to searching the web because no one is upvoting, um, Google links, right? Like an insider, an organization, one document, maybe the one you want, but it's unpopular because only one person has access to it. You, right? So it's the same thing with AI, right? And this is why something like Slack bot is so different to a chat GPT, right? Chat GPT is amazing. It's got all the world's knowledge from the internet. It's got zero of your knowledge from your company. Until you're connected to your systems. And then comes the next question. Well, who's got access to which data? What are the guard rails? And so a Slack bot tool is so much more performative inside an organization because it's grounded on all the conversations that you've all had, the documents you have access to, the emails that you have access to. And that's where value comes. So I would say there are a few areas that we're seeing where efforts can be misguided. And that's what we help our customers with, right? We don't expect to just them to buy technology. We help them along the way. We want to be their trusted advisor. And we have whole sorts of offerings to ensure that we can bring them on that journey.
Speaker 1So what I'm hearing, obviously, I heard about Slack bot. Then you mentioned chat GPT as an example. But what I'm trying to understand from the points that you're making is if you give AI, you know, and you put it into any environment that's not necessarily grounded in your contextual environment, it's not necessarily going to be set up to succeed. And you obviously would have a lot of customers that have come from a similar environment through to this. And that's what I'm trying to get at. And I think that's what I'm trying to get at.
Speaker 2By the way, it was only in English and Japanese for, you know, Japan. know you, but it wasn't in any other languages. So firstly, if you're in EMEA or Latin America, you might have some trouble, right? But secondly, that was just a repository you had to navigate your way. And you know, sometimes you found a document, sometimes you didn't. And we also had some issues. There was a lot of documents there that hadn't been refreshed or reviewed for many, many years, because every time we launch a new product, unless someone maybe flags that it's no longer relevant, the process wasn't as tight on reviewing those documents. Enter agent force, we thought let's deploy an agent on the website, completely revolutionize how an interaction might happen there, which is a natural language conversation. Just ask the agent what you're looking for. And it would use that grounded knowledge from our knowledge base of 27 years to respond to the customer. Now, that started off really, really well, but we saw some things. So the first one I highlighted just a minute ago was we noticed we had articles that we'd not shone a light on for 20 years. And they're like, hey, that's actually feeding the LLM and giving it incorrect information, which is back to that point of hallucinations, right? Saying things that are no longer in the product. So we learned that. The second thing was that, to my point earlier, it was an IT project at first, but actually the experts of delivering information to a customer were our support engineers, our customer support team. They've been doing this since the beginning. And what we realized is that actually giving them the keys to drive and tweak the agent radically improved what's called the deflection rate. Someone can answer the question themselves. In fact, we're now seeing that 84% of cases are handled autonomously. And it's because of little things. I'll give an example. Someone might have literally in IT listed that don't talk about these other brands. They're not Salesforce, but we at Salesforce connect with lots of solutions. We connect with Microsoft. We connect with SAP. And so by customers asking, how do I get my sales cloud to work with Microsoft? The old agent would have actually said, I can't answer that question. But we have a lot of solutions. And we have a lot of solutions. I can't answer that information. But actually we, you know, the support engineer would have answered that question, right? So this is where the support engineer bringing that business expertise and knowledge. And it's something that we always refer to as the humans working alongside AI, because this is not just an agent going off and self-learning doing everything all by itself. Like there's still humans in the loop there. And we believe the people that are in the loop are usually the business experts for that area, that domain. So that was a big learning for us. And lastly, you know, now this has been running for almost 18 months, over 2.2 million cases deflected is then what does that mean for the people that were doing that mundane work before? How do we reinvent their job? Because now we have all this extra time for our support engineers that were dealing with, you know, how do I turn this feature on, or I can't get this password reset working to now be able to focus on more complicated cases and provide a new level of care that we didn't before. And that requires some thinking about what that means to your point on ROI. Because that ROI is, well, if you can handle all these cases more, some people might be thinking, well, you don't need those stuff. Well, actually, no, you need them more because now they can, they can go and solve more of those cases and potentially even upsell to some of the other solutions we have. So it's a really exciting opportunity for us at Salesforce. So that's just one area. I'll pause there because there are others too.
Speaker 1Yeah, yeah, yeah. So thank you for sharing that. What I want to do is just, if we take a step back, I'm just thinking about, you know, the audience and Slackbox, Slackbot and Salesforce's offerings. Largely when, you know, I understand what AgentForce is now, because I've been, you know, accustomed to working with you guys for a little while now, and I see the power in it. But if I think about the general consumer that I deal with in the audience, we still think of Salesforce as the CRM, right? So what is the difference between AgentForce, Slackbot compared to what everyone's already got in their organization, like a Microsoft Copilot? Because that is grounded in contextual environments. So what I'm trying to understand is the audience will be listening to this and trying to understand what is different about this compared to what I've already got. That's what I'm trying to put together.
Speaker 2Yeah, no. And it's a great question because when we started Salesforce, we had one solution, which was to help salespeople automate their processes. Now we cover all elements of customer experience, right? Sales, customer service, marketing, IT. And then to your point, with the Slack acquisition a few years ago, personal productivity as well. So that's what I'm trying to put together. I guess if I was to simplify it, we think of this as being a technology stack in four levels, right? So system of context is an area where all the data is stored, and this is what is really what feeds an LLM, right? And so we've got a huge foundation there of context where we've got Informatica, we've got MuleSoft, brings in all the data across your whole enterprise, right? If you think about your system of work, this is where work gets done. So to your point, sure, you may have a tool, another tool, any, you know, either a public available tool or one from a competitor. The difference is if you're in a tool like a sales cloud or in a marketing capacity, what you want to do is you want to bring that AI into the flow of work of what you're doing for your job. You don't want some general purpose tool. And so we've enabled agentic technologies like AgentForce in each of those areas. And so that's the difference between AgentForce and just something like a co-pilot, right? But then if you go up one level above that, our AgentForce solution that's a completely different, it's apples and oranges, right? That's an agent sitting on a website, like I talked about before, our internal case, Fisher and Paykel, HypePages, that is a completely different use case. And that's not actually offered by any of those solutions you mentioned before, even the public ones, right? And that's where you saw some great stories in Melbourne of customers that are doing that as well, you know, beyond EasyPod, Kogan, we had on the main stage. So that's the difference there. If I look at Slackbot though, and this is what we call our system of engagement. There are many other tools. Firstly, I would question, and a lot of people use other tools, do they actually work with all their teams in there every single day? Like at Salesforce, Slack is adopted by every single employee. We have over 80,000 employees. And so firstly, you have a rich source of conversation there, by channel, right? And so now you have context for the agent to be even smarter than ever before. But the real unlock is that this agent that you work with, this is on a personal capacity, can now do tasks for you. So for example, a rep is about to go into a new meeting with one of the banks, right? They need to pull in all the information that has historically been recorded in Salesforce, summarize it, and then give them a game plan for the day. Slackbot can do that because it can access all the systems you're connected to, bring that, you know, all summarize it in a confident, responsible way, right? Won't get access to anything that they didn't personally have. And now you might want to say, I want to send an email or pre-briefing, ahead of that meeting, Slackbot can do that too. And then just today we actually announced you can even bring in external agents like Claude to do work inside Slackbot. So it's really become this new operating system of work. It's the new AI system of work. We like to think of it as well. And it's really in its own category.
Speaker 1There's nothing else like it. How does it work? If I'm thinking about the points of it touching everything, how do you guys handle data and privacy because it's going to have so many hands and legs into different pieces of the pie?
Speaker 2That's a great question. And it is one about differentiators because you're right. Just deploying AI. And we talked in the beginning about what, why a lot of AI production things don't become available or they fail. And it's because what we're doing and actually, you know, we probably couldn't have planned it any better if we tried. Since the beginning, Salesforce has always been about a trusted platform, right? So it's always got access controls, guard rails, governance built in for all of the systems that we offer. So it's not just, you know, we're trying to make sure that all of the systems that we have are available to our customers and which thus have employees at those organizations. When you put a layer of AI on top of that, it works really well because you don't have to reinvent these processes, whether it's SlackBot or AgentForce, think about a process like a customer service. It would just trigger an existing process of how you do a return or how you do a change of address update. If you're trying to build that from scratch, remember, AI systems are not going to be able to do that. They're not going to be able to do that at the same time you do it. They actually generate a different answer. That is a disaster if you're running financial services, if you're running health, right? You can't treat a patient differently today to tomorrow because you felt like it, right? They will expect that when you make a claim, you know, I've got a private health insurance fund. When I make a claim, I expect it to work yesterday as I did today. That's not how these systems work. They're really good at generating texts to give you an answer, but they're not so good at following processes, and they're not so good at doing that. So, if you're running a business, you can't do that. If you're running a business, you can't do that. You can't do that. You can't do that. You can't do that. You can't do that. You can't do that. You can't do that. You can't do that. You can't do that. You can't do that. You can't do that. But you can wrap it in the conversation about collecting information, how you say it back to the customer. That can be probabilistic. That's not a problem. And that's where LLMs shine. But if you try to reinvent that, you've got a lot of work ahead. And that's why we kept saying to people, it's a terrible idea to do it yourself. Firstly, people were trying to build their own models, which is a very complicated task and expensive. And now people think because there's a few vibe coding tools out there, I'm going to do it myself, a platform like a CRM or how I'm going to do it. It doesn't work that way. And I am biased. I did study computer science. I'm in CMO of Salesforce here in ANZ now, but I know how these systems work. That's not going to scale, especially not at a larger organization, definitely not a regulated one.
Speaker 1So talking about regulation, what I want to unpack next, if you don't mind, is ownership, right? So there are organizations with agents flying everywhere. Some may be on Salesforce, some could be vibe coded, whatever it is. when an agent is deployed what's your view on ownership say it does something great fantastic but if it goes rogue and it's not set up the correct way and it pushes the wrong button sends the wrong email who's accountable is it the organization is it an individual because i think every agent should have an owner but what's your view yeah
Speaker 2i know i agree with you i think we always think of and our ceo said many many times that this is the last generation of employees that are only managing other people they're going to be managing agents agents are now part of the team it's humans and agents working alongside each other and we've got over 300 agents at salesforce and they're all managed by individuals and teams and so absolutely the accountability lies with the organization that's deploying it just like they would when they launched a website back in the day right it's hard to imagine now but you could say hey i have a website and if the website's got wrong information well that's not what i say in the branch at the bank well that's not gonna fly right someone published what was on the website right and that's your website url and that's everything around your policies i say the same with agents which is why it's so important that when you deploy a tool like a gentle gay especially if it's external facing internal facing it's a different situation you're within the four walls of your company but externally facing you want to be comfortable that a someone's managing it by the way that's how you improve it as well so you've got to have ownership there the organization is comfortable with it and that's why we believe it's so important a technology like salesforce which is grounded on the context of your data your processes the governance and all the trust that we've built with organizations over the years is so important because at the end of the day you're right if you deploy something and it doesn't work the way you expected that's different to you've planned it to work a certain way with all the right thinking and it didn't work as well so i think it's important that it's strong partnerships between the vendors like ourselves the organizations go in eyes wide open and that's why we believe that Object matter experts, the humans and our AI in the loop are incredibly important too.
Speaker 1And to your point mentioning about succeeding or not, how does an agent get a definition of a success metric? Because often I see things getting deployed and you ask them, how's it performing? And I don't know, I treat agents like interns or employees, right? They've got to have KPIs and metrics that are clearly measurable, but who sets the metrics and how does an organization go about doing that? Because managing agents are going to be slightly different to managing a team of employees, right?
Speaker 2Absolutely. And look, I think it comes down to whoever the department or the owner of that agent. And I'll give an example from our world. We deployed an agent on our website called Piper, right? And the reason we call Piper is because she generates pipeline, right? She's actually an avatar that is on our website. And you can talk to Piper conversationally by voice, or you can also chat with her, right? When we deployed her, like she is able to go ahead and book meetings in our sales development team's calendars, right? She goes ahead and has a conversation with a person on our website. She shows them the website. She can navigate around, by the way, which I think is really cool. But ultimately her goal is to, hey, if you want to learn more about Salesforce, why don't I connect you with someone on our sales team, right? But she can go in and actually book a meeting. If we said just book as many meetings as possible, then our sales team are going to be taking calls on Saturday, Sunday, 1am, 3am, right? And that can't be the right metric, right? Because she thinks that she's doing a great job, but actually that's not business etiquette, right? So for us, the performance is then deemed by the person on the other side in partnership with us who own it is, is this helping them do their job, which is securing meetings with customers and creating pipeline for Salesforce, right? And I think that is no different. If we would have had a sales development rep doing that job and we were looking at the metrics they had. I think what's interesting about AI is sometimes the metrics have to be rethought because if you just literally take an old process and try to agentify it to kind of make up a word. It doesn't work. It doesn't. Yeah, you're right. It's like sometimes reinventing something that you don't need to do anymore because if the metric was, well, how many conversations get started? Well, that doesn't matter anymore because the agent's always there 24 hours a day. It's always available. So. It's more important potentially about how many get passed on, but then you've got to be careful if your goal is not actually booking meetings, but avoiding people talking to your support team, for example, that's a different metric, right? We prefer to lean in and we want to talk to our, um, our customers, but obviously we want to make sure we're talking to them when they want to be discussed. So sometimes the customer is just doing preliminary research. They're not yet ready to really go down any further. But as soon as someone said, Hey, I want to get pricing. I really want to say if this is the right solution for me. Well, we, absolutely want to connect them with a human. So it depends on the culture of the company and also what, uh, set up they have with resources on the, on the other side.
Speaker 1There's a lot of value in what you mentioned. I think there's a fair bit to unpack around metrics, right? Because if you think about, you mentioned Piper, was that his name? It's a she, yeah, Piper. She, she, she. Uh, so if you think about sales, just think about Piper and sales, traditionally how would a sales manager manage a sales team? They may manage the amount of phone calls they make. Uh, maybe their amount of appointments, right? Is what generates pipeline for Piper, but with agents, and I don't know what the model of success is, um, with Piper at the moment, but say it's booked appointments or conversions. If it's an e-commerce business, the fundamental metrics that an organization may have operated in for many years in the past, I think in many cases are going to need to evolve based on, you know, outcome based agents or ROI being modeled and achieved in different ways. So what do you think about that? Do you think organizations need to be measuring, rethinking about how they measure?
Speaker 2Absolutely. And I think that's part of the exciting part. I'm doing that within my own marketing team, but you're right. Whether it's the sales team, whether it's our customer service team. And a good example of that was if you were measuring yourself and a lot of customer service teams measure themselves on how quickly they can get rid of the customer, which is an awful idea when you think about it, if you're a customer, right? But it's a reality. Because a lot of those customer service teams are measured on how volume that they get through. Now, intro agent for servers, which is always on. You can have like my dad, he loves having a long conversation, right? He's retired. He's got plenty of time, right? He wants to feel like the company cares about him, right? And so you can do that. You can do that now, right? The agent can have an hour long conversation. And so the goal there is absolutely not about how quickly, but there it's more about customer satisfaction. You know, how they would they go and make another purchase after that, which ultimately is a better business decision. But unfortunately, the way before was due to the world of constraint meant that you had to do it that way. And so, you know, Frank, our leader of ANZ also talks about this sort of world of abundance. And my global leader, Patrick, who runs our CMO of all Salesforce, he talks about marketing makers. If you bring those two ideas together, it's basically meaning that we're unlocking this new world where now it's no longer about what we're doing, but it's also about what we're doing now. It's no longer about why you had to do it a certain way. It's like, how can you unlock your creativity around a new business problem so that you ultimately can serve your customer better? And we've been doing that since the beginning. The mission for Salesforce is allowing our customers to connect with their customers in whole new ways. You know, whether that was the introduction of the mobile phone, whether that was connected devices. Now it's a new way of dealing with potentially agentic AI. But if you bring it back to the business principles, it's the same. You know, how do you want to show up as a company? What's value for you? And how do you grow?
Speaker 1When you talked about marketing, just want to touch on that out of curiosity. How is marketing transforming for you? And what are you seeing firsthand? Like, what are the key things that are moving in that space?
Speaker 2I think a lot of marketers and, you know, we're obviously doing this here in Australia. I spent a long time in the US. When I came to the region, I was a little disappointed. A lot of marketers were very focused on advertising and branding, which is a very important part of marketing. But it's only one of the four P's of marketing. And so what I was used to working in Silicon Valley was a lot more around not only how you describe the product, the pricing, the packaging, the go-to-market. And I think what this new era presents is for marketing to get a lot closer to the business. Because what's happened in the last 20 years, we invented a lot of technology which allowed them to connect with customers digitally. But it meant that they went down rabbit holes of like performance marketing and optimizing percent here and there for A, B, C, D, and so on. And we kind of moved away a lot from the traditional strengths of marketing to be in this world of constant optimization by channel and also specialization. You basically had someone that leads the strategy for Instagram copy and versus the images for a website over here. And that actually, and this is what I'm rethinking with my team now, we're doing this globally, is that can kind of go a little bit further. Because now when you give a tool like a Slack bot to someone on my team, they can do so much more than they could before. They can pull data around historical performance. They can create copy faster than ever before. They can even create briefs for agencies without needing meetings with a whole host of other people. So it's kind of collapsing what was required to be able to do a job and exponentially allowing someone to do more but also beyond their previous lane. And so we, with this concept of. A marketing maker for us means that you're empowered. And that's not just about having everyone doing more and just more and more and more, but actually being more more creative because if you can run to the point of, you know, 95% before 10 tests and one works really well, then that means you can pivot and then change again and do a hundred tests next time, right? Whereas before you couldn't do that because of capacity and constraints. So I think it's an exciting time. And I know we're talking about marketing, but it's very similar of what I would say in other departments too, where previously people couldn't do things and now they can, which is an exciting time, but it requires a rethink.
Speaker 1Definitely requires a rethink. We'll present opportunities and challenges if we don't adapt or take the right approach there. And talking about your customers, how are you finding your customers' customers dealing with AI agents? You mentioned your dad obviously likes to have a bit of a chat, but what do you think organizations need to be thinking about before they put agents in front of their customers? And is there any guardrails that need to go in place?
Speaker 2Yeah, firstly, we always recommend that you're very transparent. Like you should never say that an agent is a human, right? I think that's a, it's a huge no-no. And so whether that's via phone or a chat bot, if you have on a website, you should always ensure that people know that they're dealing with an agent. We found that our customers are actually saying, and you know, you know, I won't name names, obviously, I don't want to overshare. It's not my information to share, but what they tell me is that when an end customer delivers, gets value, sorry, faster, they're usually very happy about that. Because if you think about it, most people are not trying to call up a call center because they want to just chat, right? They're trying to get a specific problem solved. And if you can do that faster, easier, less friction for them, most people, nine out of 10 are going to say they want that. Yes, there's the few exclusions like my dad, who's got a bit more time on their day, but most people just want to get on with their day. And if you think about that service, whether you think about trying to buy something, you know, example, you know, if you're trying to buy something and it's a Sunday, because that's when you have time, but the shop is closed, you know, that's why we have websites, but maybe the element's out of stock and you want to ask the question, or when is it going to come? That's now introduced friction. If you could get someone that could help you, whether that's an agent, or, you know, they've got a different approach to answer those questions that follow the sun, whatever it is that they do, that's a better outcome. And so we're seeing that, but to your question, guardrails absolutely need to be there because you don't want to have your doing something differently than your humans would have. And so whether it's a customer return, whether it's something that you're purchasing, it needs to follow the governance and process of the business. And that's why we're so excited with what they're offering. We have with agent force because it is grounded on all that context.
Speaker 1Have you got an example of a time where things went really wrong? Yeah, I don't
Speaker 2think at Salesforce, anything has gone off the rails, but like I mentioned, we've had a lot of learnings, right? So I mentioned the customer service example, before our, our earlier version, um, before we had Piper, where we had an agent on our website, our, um, our agent was following up with customers out of hours. And we got some feedback that, you know, I wanted you to contact me, but I didn't want it to be on my weekend or out of hours. And so for us, it was a learning that it was, you know, you mentioned interns before when you bring on an intern, they may not know that sending an email at 11 PM at night is a bad idea because maybe they did it at uni, right? But in the business context, that's not normal. When you follow up around a question that someone's asked. So we had to tweak our agent and we didn't even anticipate this to ensure that it was a bit more like the normal hours of, of work, even though it could operate all day long. And that not only includes when it talks to the customer, but when it talks to the internal individuals. So for example, when we ask an internal, a person to do something that an agent generated work, it can't expect that it would do that on a Saturday, Sunday, or immediately. So it was kind of calibrating expectations for the agent. That would be one area. I'd say that we had a lot of learnings. The other one, which I already highlighted is that a lot of it departments and bless them, you know, that was my background. And my brother was a CIO is they sometimes don't have the business context. And so knowing that you're brought in the right people from a department that understand what you're trying to achieve will ultimately accelerate the delivery, improve the ROI, and then ensure that the outcome is better, whether it's for an external customer or an internal one. Like that's a big lesson for me. And then the last one I'd say was just how you implement this technology. You have to bring people on the journey. This is a big change, right? It's probably the first time where technology hasn't just been a new advancement, but actually a fundamental rethink of how work gets done. And we've had a lot of learnings internally of how, because we know we've been on this journey now for almost two years, how you share new capability that's available, how you make people comfortable, with using it, and then how you accelerate its usage. We've had a lot of learnings there, which is why our chief people officer has kind of come up with the framework of how you roll out technology, how you get people to adopt it. That's probably my biggest learning. And, you know, you and I met last week, I'd just come out of a CMO lunch. We're 20 of the top CMOs across Australia. That was kind of where we all landed, which is how do we bring our people along in the journey? Because the question is no longer whether AI will change departments and it's just how to do it responsibly. And the good leaders I talk to are thinking a lot about how to do that. Yes, the tech is important, but also how they have the humans, you know, that the number one resource of their organization that come along with them on that journey.
Speaker 1That's a whole topic in itself. It is. So how long have you been in AI now? It's been quite a while then. You mentioned you were at Silicon Valley prior, obviously, you know, machine learning, software engineering. So you've been around a while. So there's going to be quite a bit of lessons in your career and brain and obviously the exposure at Salesforce. If you had to put yourselves in the position of the audience that listened to this show, they generally operate a business between a hundred million dollars and a billion dollars, maybe towards the two, but generally around that bracket. So let's call them the mid market. What would you say are the two or three things that would be the biggest tips or learnings from you that you would recommend to anyone at a broad level when they're approaching this shift?
Speaker 2I'd start off with, I think they're probably, if not the, maybe the second most, best segment to be able to approach AI because they're not so large that they're going to get in all this bureaucracy of who owns their budgets and all of that, but they're not so small that maybe they don't even have a little bit of wiggle room to be able to test some of this stuff. So I think they're in the right category and they're actually some of our customers that are moving the fastest, right? We're seeing incredible adoption in this segment. So that's number one. Number two is you have to be careful to balance. I'm seeing some organizations that don't even think about how they'll deploy. Like I mentioned, CEO just says, let's go do it down to analysis paralysis of, and Australia does tend to sometimes go first, but then sort of sit and wait a little bit once they've gone first, right before they take that next step. And so we're seeing really big appetites or some of the metrics I shared, 76% of Australians are very much using AI and they really want to. But there's this big percentage because their corporates haven't given them an official tool, about 56% that are using it without disclosing, right? That they're using something personal. Shadow IT. Right? Absolutely. So I think it's really important that you find and navigate the line of being progressive, not being reckless, right? And with that, that's why we tell customers that you've got a lot of your data in Salesforce that is already possible to get features turned on across sales, customer service, marketing, and that's kind of our crawling phase, right? Just turn these features on, right? They are going to ensure you get gains. You're taking advantage of this technology. If you start sort of in that walking slash jogging phase, I think customer service and having an external agent in this day and age is a no brainer. And if you don't have one, just think about the opportunity cost of someone coming to your site and having an experience that you thought was okay, but actually they've had an amazing experience. It's not just because you're in the same industry. They're comparing you. They're comparing you to their best ever experience. And we have amazing customers. Like I talked about Fisher and Pike or high pages. I can go on on on kogan.com that are now delivering these experiences that will mean that you get your, you know, outcome faster than ever before. So, and that sets an expectation now that you need to meet. It's almost like when we don't have wifi on a plane, everyone complains. It's because another airline's already got it, right? And so I would say, start looking at your customer service, but then it's a blending. If someone then starts a conversation on customer service, it could quickly become a sales opportunity for you. So you might want to bring together your head of customer service, head of marketing and sales together and go, where are the opportunities that you could be doing something that kind of blends across that customer experience that gives you value in a few areas. And then last but not least, I would say is absolutely need to make sure that you've got some trusted people in your organization that are very excited about it, that you can give them the task, right? Of sort of being the internal, we call them agent blazers, right? Which are the people that are bearing the torch, but you need to bring everyone along on this journey because ultimately it does require a whole organizational shift to really gain the value. And that's what we call an agentic enterprise. It's an enterprise that is looking across all their departments that are bringing humans and AIs across all of them, ultimately on this unified platform like a Salesforce, then sure as the guardrails are there.
Speaker 1Okay. So if I was going to forget everything you said and just remember one of those, what would it be? Well, you can go to
Speaker 2salesforce.com, but no, I joke. Look, I think get started. You're not behind. There's a lot of people that might feel anxious, but get started. You don't have to wait till all your data is perfect. There are solutions that can be deployed in days or weeks. The Piper agent we deployed at Salesforce in a week, which is incredible. Slackbot can be turned on. If you're a Slack customer, you can turn on Slackbot in 30 seconds and you can start getting value. So I'd say get started, but do it in a responsible way.
Speaker 1Okay. Okay, cool. Leonardo, the audience want to get in touch with you. I know you're big on LinkedIn, but what's the best way to get in touch with you or the team?
Speaker 2LinkedIn is a great one. I'm very active there. I think it's a great way to share with thought leadership across, you know, people that are on your audience. You can, you can also reach out to me personally as well. No, no problems there. We can share details on the pod there, email as well. But I think LinkedIn is a great one. And also I'm pretty much at every event that's got any cool people that, so, so I'll probably be around the traps if, if you've got C-suites that are on the call.
Speaker 1Awesome. Well, good connecting and good chatting and look forward to catching up with you next time. And thanks for sharing the insights
Speaker 2today. Thank you so much, Ramon. Really appreciate the opportunity. And if I can be helpful to anyone, please let me know.
Speaker 1For sure. My takeaway from today's episode is pretty simple. AI activity is not the same as AI value. A business can buy a platform, launch pilots, train people, but still be slow on the things that actually decide if technology is going to turn into ROI and actually pay off. The real work in the organization is built around the agent, ownerships, metrics, workflows, governance. So you need to be asking yourself, not just, are we using AI? But have we changed the organization enough for AI to create value within it? This is Applied AI Australia. I'm Ramon Rodriguez. And until next time, remember, you can't control the speed of change, but you can control a cookie adapt. Thank you.
PodcastTranscript AI