Leandro Perez, Senior Vice President and CMO of Salesforce ANZ, has recently joined the board of the AANA and has been at Salesforce for over 10 years including the role of VP Corporate Messaging and Content.
He talks with Anton about how humans are now working with AI agents to deliver better business, brand and customer experiences, and touches on some new ways of measuring quality and success, as well as reinforcing how critical it is to bring the marketing fundamentals to the fore rather than get lost in the tech.
You can listen to the podcast here:
Follow Managing Marketing on Soundcloud, Podbean, TuneIn, Stitcher, Spotify, Apple Podcast and Amazon Podcasts.
Well, there’s an insight, your digital labor is never going to get bored and display some slightly negative traits like my dentist.
Transcription:
Anton:
Hi, I am Anton Buchner, business director at TrinityP3 Marketing Management Consultancy. Welcome again to Managing Marketing, a weekly podcast where we discuss the issues and the big opportunities facing marketing, media, and advertising with industry thought leaders and practitioners.
Today, I’m talking with Leandro Perez, the Senior Vice President and CMO of Salesforce, ANZ.
I’m super excited to have a chat with you, Leandro, today. There’s so much going on in the world of AI. It feels like we’re in the hype phase and exploding AI phase, and I’d love to hear your views on where Salesforce is going and how the whole market is evolving.
By way of quick introduction, Leandro’s recently joined the board of the AANA he’s been at Salesforce for over 10 years, including a role of VP corporate messaging and content. I’d love to pick a bit more information about that, and has come out of a software background, so I better be careful, he probably knows a truckload more than me about the software.
He must be a good guy, as he also coaches the Forest Killarney Football Club, I believe, which was one of my former clubs when I lived there in the 90s and the early noughties. So, we might start there. Welcome to Managing Marketing Podcast, Leandro.
Leandro:
Thank you, Anton. Yeah, look exciting times we’re living in and very happy to talk about soccer as well, being born in Argentina.
Anton:
Well, I was going to start, but my team is Spurs and they recently got drubbed by Liverpool and have given them the EPL title, so let’s not even go there.
Leandro:
Oh, no. I talk South American League only or International League with Argentina. But like you said, I coach my son’s soccer team, it’s one of my passions to give back to the community and also be involved with my son’s sport. And there three games in, we’ve won all three, so looking forward to the rest of the season.
Anton:
Fantastic super start. Now, you have a really interesting background as a CMO, how do you describe your experience, what you’ve done and what you bring to the table as a CMO?
Leandro:
Yeah, and you mentioned in the intro it isn’t a traditional path. I guess some people would call it a squiggly career. I actually never sort of dreamed of being in marketing when I was little. Actually, when I was little, I thought I was going to be a lawyer. I used to watch too much TV I think, looking at all that American TV court cases, knowing that when I fast forwarded, that’s nothing like our Australian justice system at all.
But growing up, I loved technology. My brother who’s quite much older than me, he is 10 years older than me, was one of the first folks to do computer science degree. And that meant he had access to the internet at home for us much earlier than many of my peers. And it got me really excited about computing, gaming. Not much has changed with teenagers these days, but back then this was dial up internet and like bit Mac graphic games.
And so, it actually led me into a path of studying computer science myself at University of New South Wales. And I actually came out of uni as a software developer, and that’s what I started doing in my, my early career. I was coding, I was working in backend systems like Java. Now, some people on this call may not understand those things, but it was pretty hardcore technology.
And fast forward, I found myself increasingly being attracted to the business problems we were solving with the technology. And so, I won’t give you the two-hour version, I’ll give you the two minute version, but it led me through a whole series of roles of showcasing technology with a kind of role called ‘solutions consulting’ or engineering.
It’s where you are kind of a hybrid developer/salesperson where you’re showing the technology. I got closer to the technology, went into product marketing, which led me into sort of corporate communications and eventually now to the role that I’m doing now.
But I guess one of those most formative roles, which you touched on was I spent 10 years in San Francisco. And the last few was obviously with Salesforce. And in that, I actually got a really privileged position of being able to define how you talk about the Salesforce offering, which is what we call corporate messaging at Salesforce.
And leading that team meant that I was able to get for the first time, how you position an entire company’s products offering solutions. We’re in the B2B space, we’re also one of the pioneers of cloud computing. Salesforce turned 25 last year, so not a baby anymore.
And that opened my eyes up to this sort of whole new world of how you can transmit a message, not just externally, but internally.
So, we have close to 80,000 employees at Salesforce, and the mission was to define a message. And I used to work very closely with our CEO Marc Benioff, I learned a lot from him. Define how we want to talk about Salesforce, including our values, our offering.
Take that around the world through a series called World Tours which is our sort of flagship event that we do in obviously cities like Sydney, Tokyo, London, Paris. And then have an annual, like the Super Bowl of technology, which is Dreamforce, which is an event that’s in person, about 160,000 people, 10 million online. And so, I used to lead the keynote for that.
But the fun part was actually then having everyone internally also know the message. And that’s sort of where I got to know a lot more about sales leaders and executives in other parts of the business, which led to me sort of getting this opportunity to come into a region. And that’s where I landed fast forward to 2020 pre-COVID, who knew the world was about to change, but to take on this role as CMO of the Salesforce Australian, New Zealand market.
So, I’ll pause there because I covered a lot and we could go in a lot of directions.
Anton:
There’s a lot to dive into from there. But I think that’s a really interesting background because I’m sure everybody has heard the name Salesforce, that’s listening to this podcast. And different companies would’ve had different experiences. I’m sure going back in the early days it was the sales data and how do sales teams nurture and manage their leads, which we all know was fairly one dimensional. But of course, that’s completely changed.
Clients we deal with have had good or indifferent interactions with Salesforce. And of course, you know that messaging can be quite simple or straightforward. But I think for the marketers out there with the explosion of technology, it is still quite confusing to understand if I said what is Salesforce? What would you say?
Leandro:
Yeah, obviously, in this audience, I think I’m with my people here. When we grew as a company, we’ve organically developed some products, and then we’ve acquired some products. So, the offering has changed.
It’s like when McDonald’s probably started off with the Big Mac and then they added nuggets and the chicken burger and everything else, we did the same. We started in sales, like you said, and then we went to customer service. We added marketing, we developed commerce capabilities, then analytics capabilities, so you could understand everything that was going on.
And then obviously we acquired this amazing technology company called Slack, which lets you collaborate with everyone, and then MuleSoft, which meant you could integrate it all. So, I guess fast forward we went from managing the contacts that a business deals with managing the entire customer experience that a brand can have.
So, we talk about letting you connect with your customers in a whole new way. And interestingly, like in 1999 when we started, that just meant that you could deal with them through the cloud. By the way, that was a bold thing at that time, because everyone had their own little servers tucked under their desks. And the IT department would never have let anyone move that off premise. But we had to convince people to move to the cloud.
And then it meant helping people move to the digital or mobile world. So, run your business from your phone and you could be out with your customer. And now when we fast forward to today, it means interacting with people in an AI and agentic way, which I know we’re going to get into.
But it’s redefined over all those years, but really the same mission, which is how do you interact with a customer and give them an amazing experience? And that’s something we hold very true to our hearts.
Anton:
Well, let’s pick that up because the explosion of the AI world which came onto everyone’s radar a couple of years ago, really, it’s been around for years as we know, but really exploded with, I suppose, the ChatGPT generative era. We’ve moved well past that pretty quickly.
What are you seeing, just as a quick summary, and you touched on an Agentforce has just been going around the world, so what’s your view of agentic AI and where that’s going for Salesforce?
Leandro:
Yeah, I think it’s always helpful to make sure everyone’s on the same page. And you mentioned that like AI is not new. We’ve actually at Salesforce been offering AI capabilities for a decade, it’s crazy. It’s, when I started, I was part of helping launch that first AI offering. And we call it Einstein, interestingly enough, which is at the time, our CEO was like, “We should call it Einstein.”
He actually went and got the rights to use that name, and everyone was like, “No, wait what is that?” And now it’s such a brilliant name, because who was smarter than Einstein?
And fast forward to today, we have predictive capabilities in all of our products, and we do 1 trillion predictions every single week. It’s mind blowing.
But I think to your point, when you fast forward, we went from predictive to generative and generative exploded on the scene about a year or so ago right now. And yes, it was in the public domain with things like ChatGPT or Gemini and so forth.
And then that led to kind of copilots, which was this extension, which meant, oh, on the public space you have this technology, could you bring it internally and sort of give someone a little bit of a productivity hit?
But that actually to your point, is moving so fast where we’ve actually moved beyond that now. Because what we are realizing is that the great unlock with AI and this sort of generative is not giving people that 10% extra to make them a little bit better. It’s actually having agents that can do jobs entirely end to end.
And so, that’s a bigger leap and actually what we like to call what AI was meant to be, because technology for the sake of technology, and I’m biased because I come from technology.
But when you go to the business world, just getting new tech is exciting for the geeks like myself. But for the businessperson, having to deal with KPIs like whether you’re in customer service or in marketing where you’re trying to work with sales, that’s what matters. It matters that you can deliver the business outcomes. And what agents allows you to do is allows you to sort of unlock trapped capacity that you have in a business.
So, what we talk about is that it isn’t about replacing jobs. That’s where first people go is like oh no, it means that we’re going to replace their jobs.
If you actually look at Australia and New Zealand, we have a bit of a productivity crisis. We have a labor shortage, we’ve got record low unemployment, especially in skilled workers. And the only solution we’ve really come up with is, we bring in more folks, which is fine, we’ve got that, or we outsource work across the world, but it doesn’t necessarily solve for a business today, the problem they have.
And then we still have the traditional worker who’s probably losing productivity because we’ve given them all this tech and it sort of helps them in some areas. But we’ve also created a lot of busy work.
So, what is the solution to that? Well, it’s an agent that is smart, grounded in your data. So, this is not a public LLM, like a ChatGPT it actually knows your business, and that’s where Agentforce is grounded on all the things that you have in Salesforce.
So, it knows your customers, it knows the history you’ve had with them from a customer service interaction, what you’ve emailed them through a marketing campaign, how long they’ve been a customer, their preferences. And it can now start to act on your behalf with your guidance. It’s got guardrails.
And that unlocks a whole lot of value because for the first time it’s a technology that can improve the customer experience because … we’ll come back to how hard it is to deliver great experience today but actually can help you do it at a cost that is allowable for your company to do it. Because-
Anton:
I think if I jump in, that’s really interesting because I think people are leaping towards let’s take those baseline activities or baseline roles in businesses and introduce an agent. And so, that can free up, I guess the rest of the business in many ways.
But I think it’s also potentially a bit of a trap because you go some of those basic jobs or basic roles if they’re not done well, to your point about using your own data, if the data is not good data, then you’re serving up roadblocks to customers. They’re trying to find answers or trying to ask the agent answers, not quite getting the answers they’re wanting.
And then I had to pick up the phone, which I did recently, I had to pick up a phone to a five-star hotel after such a terrible experience. And so, I ultimately fed back to them and said, “You’re a five-star hotel operating like a three star.”
Leandro:
Yeah. And I can talk personally here as well, and this is where I mentioned that I think people are misguided in their fears. Like today we’ve actually created a lot of technology that makes it harder for a company to deal with their customer.
If you remember, I don’t know about you, but when I was growing up, my dad owned a corner store and my dad knew everyone that came in. He knew what they would buy, what they wouldn’t buy, like what days they would come.
If you actually ask, and we know we won’t name names, but you’ve mentioned five-star hotels, but it’s actually hard for someone at the front desk of a five-star hotel to have that same relationship with someone that walks in, of course, because they might have only been working there for six months, or maybe they only work Monday to Friday and this person comes in on a Saturday, it’s actually really hard.
Now, if you bring that to the digital world, like an e-commerce business, it’s even harder because people are interacting on Instagram and TikTok and then they’ve got conversations going through multiple chat windows. So, it’s actually become harder.
So, what this technology, the promise, and this is why we say humans and agents working together, is an agent. What it’s really good at, and it always dependent obviously on your business being set up the right way.
Which is why we feel you need a foundation of having your customer data in order and you need to have that, and we say that that should be in Salesforce. But if you have that, the agent is really good at being able to know all of that instantly when someone starts to have a conversation with the agent.
And so, I’ll talk about our own example. So, we wanted to be customer zero with this technology because we didn’t want to be out there promoting something that we didn’t believe in. So, we have a support website, and it receives 33,000 inquiries every single day.
Anton:
Just across salesforce.com, or?
Leandro:
Yeah, it’s actually help.salesforce.com. People go there, they’re existing customers and they’re looking, these are usually the people that use the platform to sort of set it up for other folks in the organization. So, they might be an administrator, or they might just be the person that’s in charge of the Salesforce environment.
And so, they go there. Historically, we have obviously support staff. We have 9,000 of them around the world, but that’s nowhere near enough people for all the customers that we have. And so, we had like a chat sort of bot that was there, but it was pre-programmed on a sort of fixed number of questions, and it was not meeting the bar.
So, we thought, why do we deploy Agentforce on our own website, on help.salesforce.com so that we can start to promote this technology.
Now, what was interesting is that of those 33,000 inquiries today, and I think we’ve been live for four months now. So, this is a pretty new technology. 84% of the inquiries can now be handled by the agent, and only 2% actually go through to the human.
And they will be either the more complicated cases or when someone specifically asks to speak to a human, which means those 9,000 people that we have, when they get that call come through, they can be so much more focused.
They’re not there about what’s in the industry called deflection where you’re trying to push them off, they actually can … it’s not about like just, “Hey, let me get back to you.” Or like, hey, push one to go to this other department. They can actually spend time to resolve that.
Or we’ve actually been able to deploy some of those support stuff to be on customer accounts to actually help with adoption. And what we really want to do is getting them successful with the technology, not necessarily helping them on just trivial stuff.
So, that’s just one example of where the humans, the agents working together and the agent has access to all of the knowledge articles that we have produced over the many years. It’s impossible for a human to know all these things.
Anton:
To know all things, yeah. Well, that’s really interesting. So, getting rid of the menial questions and freeing up the humans as you say, oh my God, I can’t believe we keep talking about it people, like real people.
Leandro:
Yes. We call it digital labor.
Anton:
Yes. I noticed. But that’s another interesting point there around are they spending more time on the phones or more time answering queries given that they’re more intensive or deep dive type questions, maybe.
Leandro:
Well, that’s interesting because I think if you look at customer service and then we’ll come to marketing because I have some use cases too there, but the metric unfortunately in this industry has become how quickly you can get that customer to end the call or the chat sort of session.
And you really shouldn’t be doing that, you should be looking at customer satisfaction. Did you answer their question? Were you able to do that quickly? And still maybe if you asked them sort of afterwards a net promoter source sort of question, would they be happy to recommend you?
And unfortunately, if you went around the world, most departments in customer service because of capacity and businesses have to run at a profit They’ve had to make it more so of how do you just deal with them and move on versus delivering an amazing experience, which is not what most businesses want to do, but it’s just a reality.
So, now when you flip that and you can give organizations a capability of delivering a better experience at a lower cost with an agent, then the humans can do what they’re best at, which is being understanding and delivering that better experience.
If you bring it to marketing, we have a similar situation and it’s not about sort of more time spent with the solving an issue. It’s actually more time spent with customers that are wanting to buy today.
So, we then took that use case, and I said, well, we run a website for Australia and New Zealand. So, we obviously have salesforce.com, which is for the Americas. I have salesforce.com/au, which is for the ANZ market.
So, it’s a similar website, but we localize it obviously with local customer stories. We have local references to our offices and all the contact information is local.
And I thought, well, we also had a chat bot there. If we had Agentforce on our website, could we actually work out a better way for the person that’s on the website to understand if they’re just navigating to understand and collect information? Or are they actually wanting to talk to someone that wants to be able to help them through with the transaction.
So, we deployed the agent there and my mission is to qualify and marketers will know this, in B2B, that person, if they are interested to pass them onto what we have a role called the sales development rep.
And that’s a person that is qualified to understand if there’s initial interest there and kind of set up a meeting with a proper account executive.
Now, the old age problem, anyone in B2B will know this is, marketing wants to generate as many leads as possible, and then sales is trying to only get the best ones, the ones that are going to close. Using an old analogy that will date me, the Glengarry Glen Ross, is only the good leads. The good leads only. And it’s hard because what is a good lead?
And so, today when someone comes to salesforce.com and they get re-routed if they’re in Australia/Au they can open a conversation with the agent and the agent has access to all the information on the website about all of our products.
And so, they can start asking questions, I’m a small business, or I’m in financial services, do you have a product for this? Or how I would do this? It’s slightly different information to what’s on support because that’s for someone that already has the product.
And then this agent can actually progressively ask questions to work out, is this someone that may actually benefit from talking to someone about a more custom solution that is relevant to their needs?
So, is there a particular offering that we might want to be talking about that isn’t necessarily generic to the product information.
And that now, can pass to the human a much warmer lead than before. And the experience is better because this person … we know no one really wants to talk to a sales rep unless they actually want to buy something.
And so, we’ve removed this friction of, well, we’re going to only give you this video or eBook-
Anton:
That’s the best.
Leandro:
Yes. We don’t want to spend time on you. And then the sales development can spend more quality time understanding the problem that the person is trying to solve. So, we’re not wasting everyone’s time. Like if we realize we’re not a good fit, or maybe for example, they’re right now are only looking to explore for the future, that’s okay, they can actually spend that time with that.
Now, I added one extra bit here just because everyone’s all talking about AI and agents, and I want to just sort of show them the art of the possible. I didn’t want to stop there with the team.
I said, well, we’ve got the agent on the website, we’ve got this human we pass to. But even sometimes the humans have issues where it’s a weekend they’re not working or it’s morning and night and maybe we’ve got some folks on staff, but the capacity is being reached.
So, dynamically now, if I can look at that and I can actually send it to a second agent, which is a sales development rep agent. And this agent’s job is to just sort of nurture this person until the human is ready to pick up.
Correct. And so, now we don’t necessarily just say, “Hey, we are going to pass you a human, expect a call in the next five business days,” because that’s a horrible experience. We can actually have this agent then can say, “Well, look when would be a good time?” Let me schedule that meeting with 9:00 AM.
Like have some of that back and forth to your point, the menial work of setting up a time to talk or even sending another material to read, that person might want to read in between. And that keeps that conversation going and it feels better on both sides.
Anton:
It’s almost a concierge style strategy injected into a lead filtering strategy too.
Leandro:
Yes. And I know a lot of people listening are going, “Oh my God, like Leandro, all these agents, what’s going on? And I think for me, I think you should think of these agents because a lot of time when people hear AI, first thing they think it’s one thing, it’s not. And when they hear agents, they hit, or maybe it’s like one master agent that can do everything.
The way we think about it at Salesforce is an agent is doing a job or a role. And so, you give that agent a specific role. We are very big on giving these agents guardrails. So, what it can’t do, so if we don’t want it handling pricing questions, you tell it that and that’s when the job ends for the agent.
But I think what is important for people to understand is this is the reality of a new world, and it means that us as humans, we are now going to be managing humans and digital labor. And it’s no different to you might have been managing a outsource team in another world as an agency. It’s kind of the same relationship.
You have some expectations, you kind of ask it to do something. You expect an output. If it’s not, you need to give feedback. It’s not that they necessarily are doing everything you did because they’re doing part of something or different.
And so, it just creates a different way of getting the work done, but allows you, and this is what I’m most excited about with my team, to focus on the true work of marketing, which is the creativity, the strategy, working with our stakeholders. So, yeah, I think it’s an exciting time.
Anton:
Yeah. I think it’s a good way to put it because we’ve moved on from, I guess the monolithic world, where it was, IT handled it all. And we’ll just put in something, I won’t name all the other techs but put in something and that’s going to solve all our ills, which we know never worked and was never going to work.
The data quality was poor, the systems didn’t talk to each other. And then suddenly we end up with this spaghetti of tech.
So, that idea of individual agents solving individual, I guess journey related challenges or issue related challenges or output related challenges, whatever they end up being is a good way of thinking about it.
And I guess that’s then the role of marketers. And maybe it’s bringing marketing and sales and it much closer together to say, what are our biggest issues we need to handle here? What are the biggest customer experience challenges we’d like to put front and center.
Leandro:
Yeah. I want to make sure people probably listening to this want to get some takeaways as well here. Like it’s not all the first time you put in actually, it all works exactly how you intended. So, let me share a few sort of learnings-
Anton:
I was going to ask you, can you give us the train wreck, or some of the challenges that you faced?
Leandro:
Yeah, no train wrecks, but it’s all incremental. It’s like when you bring in a new intern or a new employee, you expect them to not know everything or to make a few mistakes. So, the way we think about agents is you put an agent to work, they’re just like a new employee, you’ve got to give them the information for them to be able to do their job.
And so, if we look at that help.salesforce.com support use case, we gave it the training of all our knowledge articles. Two issues came up at first, was, we had hundreds of thousands of knowledge articles that, by the way, none of the human agents had ever read 10% of it. They’d kind of read the newer stuff. But because we fed it all to the agent, it actually started going back and using some outdated materials.
So, it actually helped us identify that we had some outdated knowledge articles that we needed to curl. So, that was an interesting process because we were like why are people saying that this, UI thing is not here anymore? And it’s like oh, our knowledge articles has this archive of things we didn’t even know existed anymore. So, that was number one.
And then number two was like all good training. When you bring someone onto a company, you want them to represent your brand as best as possible. We had actually instructed the agent. And I should add here, the agent by the way, you don’t code it, you just tell it in natural language.
You tell it “You are going to be a customer support rep. You are going to be courteous. You will use these things as the knowledge and articles. If you don’t know the answer, don’t make it up, we don’t want hallucinations. This is when you should pass on over to a human.”
And one of the things we had asked it to do was if a competitor is mentioned, just sort of don’t answer that question. Sort of just say — because we anticipated that people would be trying to ask the agent sort of to cheat it into thinking, is this product better than something our competitors offer?
Which a human would not do in a job, because they’re employed by the company. But we wanted to tell the agent this. Now, what we didn’t anticipate was that we integrate with all the technology. We integrate with Microsoft, Google, Amazon, like all the technology because you have to in this day and age.
And so, when customers were asking how Salesforce would work with some of those platforms, the agent would be like, “Can’t respond.”
And so, when we realized this, all we needed to do was we added another sentence and said, “You can react, but just always make sure you put Salesforce in a positive tone.” Which is how we would anticipate an employee would do that. It’s not about saying anything in factual, but which then meant it opened up to the ability to answer all the questions on connectivity.
But if someone was asking it a comparison question, it would say, “Well, here’s what we do.” Rather than a negative of here’s what we can’t do or comparison. So, that was a really interesting flick of the mind.
On the marketing side, we deployed that SDR agent I talked about and like all eager new hires. So, when you hire an intern that really eager to do things they may be not so good at, what we realized is that the agent was too eager to respond to the customer.
So, basically it would send an email and if I hadn’t heard back in 30 minutes, it was so eager. It was like, “Hey, just checking in again, have you read the PDF or the document I sent you, or here’s another video.”
And so, we just had to tell it to tone down a little bit its level of excitement. It was too helpful. It was like too aggressive of like just chasing up and even including over the weekend and stuff. So, we just had to give it some guardrails of, hey, if you haven’t heard back, like this is an expected sort of SLA of when a busy CIO might respond. And so, that was interesting, we didn’t anticipate those things.
But to your point on working, this was a good partnership with our IT department. Obviously, our sales team needed to be part of the picture because we have essentially changed the relationship between marketing and sales there.
So, we are reducing the number of leads, which in marketing is a big no-no if you’re in B2B, right? But we are improving the quality, which is a good thing. But you know any CMOs in the B2B space, that’s a hot topic? Because if the quality’s not better and you’ve reduced leads, you are in the sin bin.
Anton:
You’re right. Well, I think it’s a big risk is taking on that, now it’s going to come to trust, but the trust of everyone saying we’re going to deliver higher quality leads. Okay, yeah, sure. Sales came through them all and say, “No, they were all rubbish.”
But where does the trust come from in your experience, or maybe a case study examples, but trusting the quality that’s getting passed on. So, lower volume, how are you seeing the quality and the trust being rebuilt or built?
Leandro:
I’ll answer and then I’ll push to a point on some other examples because obviously I’ve been talking about ours so that I can speak sort of from the heart. But we’ve got thousands of customers all around the world now on Agentforce across industries and different size organizations from small to large.
I think trust is interesting because there’s a couple things. One, you need to be able to trust the technology is doing the right thing.
And so, a big part of when agents sort of first hit the scene, we were talking about the trusted AI layer that we’ve implemented, and we call the trusted Einstein layer. Because when you have an LLM and LLM is a new piece of technology. So, firstly, most people are still trying to get to understand what that means.
But a lot of companies were, either trying to build their own LLM, which I wouldn’t recommend. DIY, AI is very fraught and hard and many bodies on that path, I would say eloquently. But then the next bit was too much trust, like send it all of our customer into information and no thought of PII or anything like that.
So, what we wanted to implement was when you leverage something like Agentforce, firstly you can choose your own models, but it’s based on your existing customer data that you would have in Salesforce or in other sources that we can access. But you are not sharing that with the LLM unless you want to. So, that’s number one.
And there has to be some things like first you need to mask some things, when you send personal information to the LLM, it doesn’t need phone numbers, it doesn’t need ATO things, it doesn’t need medical records. It just might need a first name to address a human or that actually could be added on the way back. So, there is some things that we’ve implemented to make Agentforce safer from that point of view.
Also, LLMs might have some hallucinations or toxicity that can introduce in the language because they’ve been trained on the open internet. So, they’re not necessarily trained on everything that’s ever been said. It’s only what’s been said in a written or a recorded form on the internet.
So, this layer is very important because we sell to financial services organizations, healthcare organizations, and they want to ensure that what they’re using can be audited and they can sort of dial up or dial down some of these filters.
The other point of view from a trust point of view is does it deliver results in a way you would expect. So, that’s probably where, beyond my examples, I’ll share a few.
So, a great example is just across the ditch in New Zealand with Fisher & Paykel they’re one of our early trailblazers with many of our technologies. But if you know anything about them, they deliver premium appliances.
So, they’re not one that wants to damage their brand when someone calls up for customer support when they’re dealing with an agent because they’re all about a premium experience.
So, for them, what was really important was they realized that when someone calls up about one of their appliances, it might be that the appliances is not working and they don’t have days to potentially solve it. They need to kind of answer it quickly.
Like their oven is having an issue or if their fridge, God forbid at my house, if the fridge was out for 48 hours, I know my teenage kids would be like, “What is going on? It’s like end of the world.”
So, they knew that they couldn’t handle all the inquiries as quickly as possible with the human agent. So, where their agent is deployed is trying to see if there’s any simple resolutions that can be handled.
And they’ve actually now are handling 66% of their inquiries through the agent. And that’s right now via text, as in chat. The vision that is the next step is voice, so which is going to be really cool actually.
And I demoed some of that technology at a recent event and everyone’s mind was blown about it. It’s talking almost to a normal agent but being able to access the data that Fisher & Paykel will have about your purchases.
And a key thing for them is if they can’t resolve it straight away, is they want to send a filled technician out to your home or work out how you can get a return, or a warranty done. And that usually takes a bit of back and forth on schedules. And that’s a job that no one wants to do. It’s like let me look at my calendar, let me look at your availability. So, that’s Fisher & Paykel, a great example there.
Another one that I love is hipages. So, hipages has … and I don’t know where people are listening into from around the world, but obviously if you’ve ever wanting to book a tradie to your home, you go to hipages because I don’t know the local electrician in my area that’s available tomorrow to do something.
So, you go to hipages, you get a few folks that come back to you as potential that could come to your home.
Now, what hipages’ challenge was is their tradies are kind of this huge pool of people that come in and out of the industry and they want to verify that these people have the right licenses and the right applications to do the work because if they send someone to your home, an electrician supposedly, and they don’t have the right licenses, then that’s not a good situation.
So, they use Asian force to verify the tradies on the platform. So, what used to take many weeks to verify tradie, now Agentforce asks, I need this license. Do you have this? Or I checked it, and your date is expired. Can you provide another copy? All that kind of triage to get a tradie as quickly as possible onto the platform, which has really improved the ability for them to get them on.
So, they’re are just a couple of sort of local examples that are across industry and different use cases.
Anton:
So, on both of those they’re both good ones, really good ones. The thought in my mind comes to, we’re in a completely new era. You wind back to the start of this conversation, your father running a corner store.
You know people, your database is in your head, and you can have a conversation that can meander into all sorts of areas. Sometimes about your partner or about the kids or about someone’s birthday. And by the way, I want some fruit.
So, I wonder whether this modern age though, are we getting, and I’m going to use the word robotic, but are we getting too robotic in dealing or in thinking about individual agents and just solving a problem?
So, why I called in? or why I am needing some information? And where’s the humanity … and again, from a brand or a marketer’s perspective, it’s a fine line, I think now with levers to go, “Okay, I get the sales angle, and I get the lead triage, but I also want some humanity and values all to shine through from the brand’s perspective, which is quite human.
And we’ve talked a lot in marketing about bringing the emotional side of marketing versus the functional side. Decades of we’re trying to drive emotional attachment.
Wonder what’s your thought there, are we going too robotic or are we teaching the robots to become a bit more empathetic and a bit more emotional in their response?
Leandro:
Yeah, it’s interesting. I could take this in so many directions. I think number one is an agent is an extension of your brand in that role. And like interestingly when we deployed an agent for Gucci, like they’re the top of the top of premium white glove service.
And that Gucci, they’ve actually got this massive brand guideline of how you should interact with a customer, in tone and the types of things you ask, or you respond with. And that was fed in to the Guccified tone of the Gucci agent.
And the Gucci agent actually is not trying to sell you anything because they know the customer they have is not … they’re not trying to have any sales or it’s really just about offering this warmth of being in the Gucci family, so to speak.
So, that agent absolutely is very different to the one that if you are going to be in a retail cutthroat 1% margin organization where you’re just trying to get me my response right now for my refund or whatever.
So, I do think you can bring your brand to it, and that’s why I’m so passionate about marketing and CMOs being part of this journey and having a seat at the executive table because I truly believe that CMOs are the closest to the customer or the marketing team through extension of the CMO.
And I worry that if it drives this project, that will go the path that you talk about. Which is put in an agent to handle a task but not think about the customer experience, not think about it end to end as a customer journey.
So, that’s why I like to talk about CMOs. So, firstly, you can’t bury your head at the sand on this stuff, you really need to be part of the journey. Even if the agent is not deployed. Like if it’s a customer service agent, ultimately that’s a customer experience. And we know that support can become sales now or sales can become support.
So, the lines are blending, yeah, of like we created these internal departments, and they get handled in these channels, that’s not how a customer thinks.
So, I feel a job of a CMO now is to think about the technology that’s going to be interfacing just like it was when social came out or when e-commerce came out. We became part of these conversations some later than they should have been.
So, if you are to seat at that table, then you can start thinking about, well how do we want our brand to be interacting?
And you are right, we want the empathy to come through. And something I’ve actually learned is agents can actually be more empathetic than a human because I don’t know about you, but every time I call my insurance provider, that person is dealing all day long with someone either having a claim, which is a terrible situation, or someone doing a renewal and they’re trying to lower their price.
There’s very few positive situations in that relationship. And a human, everyone’s human, you’re going to get tired of the same conversation, whereas an agent, every new conversation and they only get better at doing the conversation with learning, they just approach it sort of in a way that you’ve asked it to. So, be empathetic, be very polite, it doesn’t get tired of doing that.
Anton:
Well, there’s an insight, your digital labor is never going to get bored and display some slightly negative traits like my dentist lady, she’s got the world on her shoulders. I just want to go and get into my appointment.
Leandro:
Correct.
Anton:
She’s tapping away on a computer, not even looking up yet. It’s like how about you serve the customer first and deal with whatever you deal with?
Leandro:
But I also feel as humans, it’s important for us. I always teach my kids, and I say this to our customers, that we also have to be polite to the digital labor and the agents because they’re learning on how we interact with them too.
And so, if every customer that rings up or is on a chat session or is angry and uses bad tone, they’re going to think that we’re always coming in hostile.
And so, I think we’ll allow potentially a world where we can all realize that, because we’ve made the world more complex with all these systems and easier to get to technology, but now this is potentially a technology that allows us to be a bit more human. So, that’s number one.
Number two, I would say is that when you are dealing with this new world, we must not forget that us as consumers may start to get this technology ourselves.
So, Leandro might have an agent that helps him with booking his next travel or with dealing with the next insurance quote for the home. And so, we’ve got this world that it’s not just brands that are in this area, it’s humans and consumers are going this world too.
So, we need to be careful that if brands are … so my message number one to other folks listening, if you’re not on this journey, you may not have a choice because I Leandro, as a customer of yours might have an agent coming to your call center soon.
And so, you want to be prepared for that, because the volume of that could be hard. I might have my agent call in every day to get a better quote, for example if I’m not happy with it.
And then number two is, if your brand is not dealing in this way where it’s more personal, more empathetic, I Leandro might tell my agent, “I don’t ever want to see that brand in my world.” Which is we’re potentially creating worlds of where the agent kind of dictates what I see.
So, it’s actually interesting, it’s this both side equation and ultimately, I think it will lead for a better outcome for all of us. But you need to be participating in this otherwise you will be left behind.
Anton:
Yeah, I think it’s a really good message. CMOs are getting hammered in most articles you read in these days. And it feels like a drag and a heaviness on all poor CMOs that are dealing with all these different directions. But there’s a lovely insight there to say actually there has never been a better time to lift up above this.
Second, those brand guardrails, brand experience, customer experience, and make sure that is baked into any future activity you do, which links to the IT teams and the sales teams. So, yeah, it’s a good opportunity.
I guess underpinning all of this, and you mentioned a little bit earlier, I was going to touch on it. You didn’t recommend going DIY with an LLM. I get that, but I think what we’re seeing is the advancement of almost no code, low code that I can develop anything.
And you said, yeah, I could develop my own Anton agent, et cetera. Whether I use trusted sources or whether I just build something.
I worry personally about this massive tale of stuff that’s going to get developed. I wonder on your thoughts there and I guess some people listening could go, “Yeah, I don’t need Salesforce, I could go and build it myself.” And maybe it’s the size of company, larger companies of course probably wouldn’t the smaller companies, potentially even sole traders could say, “Oh, well I’ll just build my own, give it a go.”
What are your thoughts there around the explosion that could happen here? Because it could be an absolute train wreck as everyone starts developing stuff.
Leandro:
Yeah, it’s interesting. I think there’s two angles to the DIY. I think the first angle was, and this was more the larger companies, not the small, but the larger companies immediately thought large language models are an asset, I should build my own version so that I’ve not locked in. I’ve got my own proprietary LLM.
Now, what we realized through the industry is that LLMs will largely be probably commoditized in the future because most of them have been trained on all the texts that are available on the internet. And yes, they’re fine tuning them here and there, by the way, they cost hundreds of millions if not billions of dollars to create.
The layer on top is the reasoning level, which is probably where the next evolution is. Kind of like what I was telling you about, like how it thinks, like does an agent react in this way or what it can do and what it can have access guardrails. That’s probably the actual IP and most companies to do that, you need engineering teams and so forth.
So, unless you want to become a software developer, I wouldn’t recommend … as in a whole company becomes an engineering software development company, I wouldn’t recommend that.
What you are describing on vibe coding, which is the term that’s used for that at the moment, is interesting because you are right, anyone can now go, and this isn’t an LLM, but using LLMS to go and create software that solves a task is going to explode.
And in fact, in my personal space, like I’m using that myself because I love this stuff where I’m like, “Hey, can I use it to help me with my finances? Let me build a thing.” But when you bring it to a company, the reality is why we’ve been on this digital transformation journey for the last 20 years is that the world is not as simple as it seems at first.
And let’s not say small example, but you’re a midsize bank, you have a lot of regulations that you have to deal with, you can’t show Leandro’s accounts to other customers. That’s a big no-no.
Leandro’s account even shouldn’t be sort of shown to everyone at the bank. Just because I’m a customer of yours doesn’t mean that everyone at the bank should just be able to go in and see all my transactions.
There’s a lot of sort of complexity around the metadata who should see it, privacy regulations and so forth. And the risk is you go in — if internally at a company you’re like let me develop an app to deal with a customer, you have no idea of how it’s created, because it just generates code. And usually when you’re developing that, it will be created with the privacy restrictions that you have as an owner in that company.
So, you are going to give all your permissions to everyone at the company because it doesn’t know that everyone has different permissions.
So, there’s just a lot of risk that you are introducing that, A it will overshare, or it would get it wrong, and you wouldn’t even know, because you’re just interacting with it.
Now, that was a bank, let’s go even a simpler use case like travel and expenses. In any company, there’s someone that probably submits expenses and there’s probably an accounting team that needs to be able to verify and use it. Again, do you want the CEO’s expenses being shown to everyone?
There’s just so many things that we’ve built up for a reason at a company, and if you’re not doing that on a foundation of trusted data with some guardrails, I do feel it’s the wild, wild west and you’re just setting yourself up for future disaster, in my opinion. And by the way, I’m the biggest proponent of innovation and I tell my team to try some of these things.
I just don’t think it’s at that brand level, like for your whole company that I feel you want to make sure … and it doesn’t have to be Salesforce, but you have your data in a trusted place. You have some governance and guardrails around it that is kind of usually what the IT or the CIO will be very adamant about.
What you want is then democratize how you access that data and yes, maybe you’ve got some vibe coding happening in the edges, but where I’ll come back to is the CMO hat on. I’ll take off my Salesforce setup, but I’ll put my CMO hat on. Do you really want 50 departments building agents, talking to a customer in 50 different ways? You probably don’t.
Anton:
You’ll lose control of your brand or your tone, et cetera.
Leandro:
It’s back to when, if you go to a brand, when we introduced websites, every department, one of their own externally facing website for their product, they’re like really? If I’m a customer, I’m like oh which … if I go to Tesla and they’ve got a different department for the three versus the why versus the when I go to the customer support.
I’m like you don’t want that as it, you want to deal with one organization that feels homogenous, that has a repeatable way. And that can only be done with some sort of level of sophistication and agreement across your brand. So, I think it’s interesting, these are all new conversations, we haven’t had to have these before.
Anton:
They are, but I think some things haven’t changed. Your point there about the website era, the late 90s, early 2000s, just build another site, another site, another site, and there’s a whole bloody universe. Then we got to apps, building-
Leandro:
Build your own app.
Anton:
So, some things haven’t changed. The logic of what you’re saying, need a brand strategy that hasn’t changed since day one. But interpreting that brand, I like the idea of the guardrails. We’ve seen it in brand documents, brand bibles, it’s all been sitting there, but how does that come to life in this digital or agentic era? Where I think, as you say, I’m on the same side, it’s exciting, but just set some guardrails.
I like your Gucci example. So, it’s not necessarily about sales, it might be about just a customer experience and being, and talking about the Gucci world, knowing the Gucci customer.
As we all know, every customer is slightly different, Anton calling into a hotel might be different to Leandro calling into a hotel if you want to have a conversation, otherwise I just want to book it straight away and have it seamless and getting get out.
But yeah, it’s exciting times, I think one of the key takeaways for me is that individual agentic solution and how those agents are going to talk together much more powerfully in the future as well.
So, whilst you can set up what we are probably saying satellites, but different agents solving different challenges the human side of everything as you said. But as agents start to talk to agents, I guess potentially pass a challenge over or pass an opportunity over or whatever it may be is exciting times.
Leandro:
I guess if to sort of summarize, I think firstly everyone should be looking at this space right now. I think it’s going to affect all of us, all industries. So, you need to be curious and exploring.
Obviously, we have a solution at Salesforce with Agentforce, and we feel what the way we’re approaching it is in a trusted way, and it’s based on good principles and guardrails like you said. But we know there are other providers.
I would say if you are a CMO, try to get a seat at the table in this conversation. I’m a big believer there that you should be having those conversations, whether with the CEO or the board and making sure you have a point of view there. But you got to bring your teams along on this journey as well. So, it’s not just about the leaders.
And yeah, if you’re going to muck around with some vibe coding or something on the team, that’s fine, as long as the expectation is this is not going to go and be your enterprise wide strategy. But encourage that little bit of innovation because that’s what’s going to lead to an idea of how you might want to do something. It’s not necessarily going to be the implementation of it, but an idea of, how you might do it.
But overall, I think it’s an exciting time. You are right, some of the old problems have come back and you don’t want to forget those lessons. But it’s an exciting time to be in technology, to be in marketing in my opinion, and don’t lose the grassroots of what marketing is all about, which is delivering an amazing brand experience. And that’s not changing.
Anton:
No. And maybe that helps CMOs because we’ve heard it that yes, get a seat at the table, talk to the C-suite present at the board. But there was a sense of nervousness, a bit like the digital era. Having brand CMOs talk digital was not a comfortable space to be, so therefore, for many didn’t for a period of time. And then of course, exploded.
I wonder if this, what are your thoughts around the CMO who’s not as skilled or forward leaning into AI yet? How do they approach it to either get skilled up or are starting to approach conversations either at a board level or a C-suite level?
Leandro:
The good news is that there’s not that many experts in this space at the moment, there’s a lot of people that say they’re experts, but there’s not that many actually.
So, especially locally, I work globally, and yes, there’s quite a few around the world, but like locally, it’s far and few between, especially in what we’re talking about. There might be some people that know AI like machine learning and the actual technology really well, but what we’re talking about translating that to business, very few. So, that’s number one.
Number two, it’s a fast-moving environment, so if you just even just get curious in learning with what’s happened the last six months, you’re not that far behind.
So, I think that’s a good time to be getting in. And I welcome anyone, we’ve got Trailhead, which is our online learning platform, you can go there and learn Salesforce and non-Salesforce, sort of principles about what’s happening. So, I’d recommend people to get started there if they don’t have any other things they’ve already looked at.
And then bring it back to the principles because what we’re talking about yes, is the technology, but if you apply it back to what marketing should have own and a lot of people here in Australia and New Zealand in particular think about marketing as the promotion of the four P’s of marketing, but it’s also the positioning and the packaging and the pricing, and that should be no other responsibility of other teams.
And so, leaning into that, and in fact I won’t name names, but I was talking to probably one of the best CMOs in our country representing one of the largest telcos that we have. And big brand guy. And he actually loved what I was saying because he mentioned that this makes brand more important than ever.
Anton:
Exactly, never been a better time to be a strong brand marketer.
Leandro:
Yeah. If you don’t have a strong brand, A the technology might level you so you all sound the same, but also you could easily be forgotten or not be considered.
And so, more important than ever. So, you connect that with … you don’t have to be the expert in AI, but where you want to be an expert is in the potential and how you want it to be, and that’s the strategy.
So, I would say don’t fear, obviously all those situations can be a little bit scary talking to a board or CEO, but they’re there to also navigate this complex time. So, I actually get the privilege of speaking to a lot of CEOs, CIOs, boards as part of my role. They’re all right now hungry to work out what they’re going to do. They haven’t figured it out.
And so, having people at the company that are willing to raise their hand, they have a point of view they’ve gone and explored, whether it’s talking to us or other organizations to help them chart this course, they’re going to grab onto those track because they need those people.
Anton:
That might be a great place to end it. Lean in, get involved, skill up and keep conversing. So, if you want to touch base with Leandro, I’m sure he’ll be able to share some wisdom with you. But really, thanks for your time today, Leandro.
Leandro:
My pleasure, Anton. I’m always available on LinkedIn for folks who want to connect there. And what a great conversation. I feel, maybe we’ll look back in 12 months and go, “Wow, we had some things right, some things wrong.” But I think that we’re on the right track in my opinion, and it’ll be fascinating.
We always underestimate what’s going to happen in 10 years. So, we overestimate what we can do in one year, but I feel if we look back, we’re just here at the beginning of a new generation. So, very exciting times.
Anton:
Excellent. Well, I’ll pick that up in 12 months and we’ll have a chat and see where it’s gone. I’ve taken one thought out of this, the average handling time, AHT, we’re going to rename that, the agentic handling time.
Leandro:
Oh, I like that.
Anton:
Hopefully that’s gone up because it’s higher quality and longer conversations on the phones. Leandro, I’ll have to challenge you to take that through as a measure to all contact centers and all marketers herewith.
Leandro:
I love that. I thought you were going with soccer there for a second. How long does it take my son to release the ball or anyone on his team, which you know hard in the teenage years, but I love that. I love that.
Anton:
If you’ve enjoyed this episode of the Managing Marketing Podcast, then please either like, review or share it to help spread the words of wisdom. Bye for now and thanks Leandro again.
Leandro:
Thanks Anton.
