Tiny Mic: Why Your AI Strategy is Probably All Wrong (and How to Fix It)

Be less concerned about the how. A lot of times people are like, I want this what, but here's how I want to do it. And this agentic technology is just so new and changing so rapidly that if you fall in love with the how, you're probably going to get left behind. Find out the what and be flexible on the how, and then just go for it. Just get out there and start practicing.

Hi, I'm Erin Uhls, SVP of Strategic Events and brand Experiences, here live from Cannes Lyon on the beach with Joe.

Ensurillo, Chief Digital Officer.

Thank you, Joe, for being here with us today. We're super excited to sit down with you and get your honest read on what you're seeing here at this event. I know you're talking to a lot of executives. You also have a lot of contacts here from your very amazing background. So we're just getting started here in Cannes, but I know you have experience here as well so far this year. Any big takeaways or things that are jumping out at you?

Well, I mean, I definitely think, you know, it sounds a little bit trite, but it's definitely the year of data and the year of AI and so those two themes. You can't walk past anybody's beach, go to any hotel without hearing people talk about that. And obviously, that's where we're very focused as well. So we're definitely there in the pack, and I think in a lot of ways, leading the pack.

I was going to ask a little bit more about that. You're checking in with different brands as you explore and have different meetings. How do you see Salesforce in the way we're sitting in, the conversations we're having about AI as compared to those that surround us?

Yeah. You know, it's funny because I think, you know, you used to express, like, how far companies were in terms of years, and now I think it's months. When we look at where we are with AgentForce, you know, we've come such a long way since the debut of Agent Force, and we're really starting to get good success stories and good customers that are taking real value, making great experiences. But I think when you look at the spectrum of people I've talked to, it's just all along the mix, there are some folks that are terrified and just getting started, and there's some folks that are going to finishing school on their first sort of set of agents and things like that. And I think we're definitely towards the front of it because I think we've really gotten to a Point where we can now demonstrate a lot of value. And we could do so very quickly, especially for customers that have our other clouds that sort of just turn it on and really start to experiment.

Yeah, the barrier of entry is a little bit lower for those that are really ready for it. And then it sounds like what you're saying is that we're starting to establish ourselves truly as a trusted partner that can help guide perhaps those that are more skeptical and need a little bit of support in those early conversations.

Yeah, I mean, I think, you know, a lot of folks that are doing this for a big brand or company or a small brand or company, they're trying to figure out what their path is and to chart it. And so certainly for folks that we already have relationships with, we're a natural trusted partner. We've been with them for a long time, they've grown with us, they've seen us deliver real value. And so I think we're a great on ramp into it. But I also think that we have the potential. I'm seeing signs of that early signs of it where we're starting to break out into companies that maybe haven't thought about us first as an agenta company. But I think the message is starting to get through that we have a real solution here, you can put it to work and you know, it's got a lot of power. And so I wouldn't be surprised if we wind up with companies that start agentic at some point in time and then start to experience the rest of our clouds as opposed to folks that already have our clouds moving into the agentic space.

Incredible. You've worked at a lot of incredible brands, including Disney and SiriusXM. You were a customer of Salesforce much before you joined us, which we're very excited to have. You tell me what have been a couple things that are surprising to you in that shift and how do you take that customer perspective and, and apply it in the conversations you're having?

Well, I certainly think that all customers think they're the most important customer. And so one of the amazing things that I've seen is just the breadth of customers we have and the difference in customers. So we just have an incredibly huge, huge base and an incredibly diverse base of customers. Sometimes the things outside in that seemed easy, you can understand why they're a little bit more difficult now. But I also think that one of the things that I do bring to the table that resonates with folks that I'm talking to out here is really that practitioner mindset having Done it, having operated these things, understanding what the problems are and having a good feel for the rest of the ecosystem and how it plays into it. And I think that's a big part of what I'm here to do, is to provide a sounding board. And in some cases it's just strategy and things like that that people are still trying to sort through with something like data cloud, where you can now fuse the technology across a bunch of organizations that may be very diverse in the past, different instances, different clouds, things like that. Even for folks that are already dyed in the wool salesforce customers, it does bring up operational challenges. And so going back to some of the large companies that I've worked for, especially when you can say, well, this is the way that we approach this, maybe that would work for you, I think is helpful. And that gets to the trusted partner thing that you mentioned before, which is us being a trusted partner has to be somebody who's not just always in a selling motion, but really in a success motion the entire time. And part of that success motion is bringing some outside perspective.

There needs to be a thoughtful conversation in supporting them.

Yeah, I think that's right. You know, the AI stuff is really interesting because you could tell even from some of the technologists that in some cases they look at their other folks as barriers, you know, that they're not keeping up the ex legal team or the privacy team or whomever is not keeping up with pace of change of the technology. And then for other folks, they're actually happy about that because they're a little terrified about where it's going too. And they're like, well, okay, it's okay that we take a little bit more conservative cut because I'm still learning too. But I think like I said, it goes into that spectrum of where our customers are and we just need to meet them where they're at. Some of those things may be like, hey, let's try a pilot. Let's try something that's a little bit less ambitious just to get the reps in and feel comfortable with the technology. And then when we see that we can actually alter the trajectory of the business, we can bend the curve in a positive direction. Then we can go back and keep adding and adding use cases and adding consumption. And I think especially for us, when we start doing things like Data Cloud and AgentForce, which are really consumption sales, we want to spin those dials as fast as possible, obviously for ourselves, but we also want our customers to have some control. And so that really does allow people to ramp in and try to do things and try to experiment and just get out there and get after it and just get some reps and you know, the version two is going to be better than version one, version three will be better than version two. And that continuous improvement cycle, I think is something that's essential in the agentic world, especially because it is not a fixed point in time. There's no such thing as ship it and it's done. It's ship version one, then ship version two, then ship version three.

Right. And the pace of innovation is outpacing any innovation that's ever come before. So we have to stay fresh and be paying attention to what's happening for each industry and how to remain relevant there.

Absolutely.

Now you talk a little bit about hero agents. Can you tell me what is a hero agent?

Yeah. What I've seen both for ourselves internally in my previous experience with the Gentix side and what I'm seeing with customers is the customers that have a really, really crisp idea of the what they want to do are the ones that are successful. So really defining what this agent role, don't make up a role that's like, well, we have this problem where these four departments don't quite work together, so let's actually put an agent there and the agent will solve all the problems is not a great way to go. So it's really saying, okay, what's the subset of the sales, let's say associate's job and what do they do every day that we can automate? And then making an agent that's very directed and very specific to that use case and then again continuing to iterate and improve that agent over time, that's what's actually delivering real results. And so sort of narrowing that scope down to something that's achievable and then something you can iterate on, I think is absolutely essential. Eventually we will get to multi agent coordinations, all those other sorts of things that we get coming down the pike. But if you don't start with a real focus on the what, you're never going to get there. And I also think that's equally important is to be less concerned about the how. So a lot of times people are like, I want this what, but here's how I want to do it. And this agentic technology as we talked about is just so new and changing so rapidly that if you fall in love with the how, you're probably going to get left behind. So it's really about find out the what and be flexible on the how, and then just go for it. Just get out there and start practicing.

We saw that with Dreamforce when we launched Ask Agent Force. We just knew in a big picture idea, here's what we want, but really, really didn't think the technology could build it. But then we brought in the right technologists and really were blown away at what it delivered and how much it improved the experience. And then we were able to roll that out to Davos and customers anywhere can now do the same thing. So I think it's a good example of we didn't know how it was going to happen, but we had the idea and I think starting there is a great place to start. How do you know when you have a great hero agent that you can go after? Like, what are the criteria that you see of like, it solves these things or it's defined in this way that might be helpful for our team at home to think about and when they talk to customers. How do we get to your hero agent that we should start with?

I think a lot of it is measurement, right? If you can't measure it, it sort of can't really be iterated on coming up with like, what is the thing that this agent's going to do and how are you going to know that it's doing it? So a lot of times when we talk about something, let's say like sales automation, what we're really looking for is a reduction in time that people are doing or an increase in throughput of the deals they're able to work at the same time. But really trying to figure out what those KPIs are. And then it's something that you can iterate on and just continue to refine and torque on and get that number up or down, depending on which direction you want it to go, without the sense of measurement, then it just becomes a very subjective, well, do we think this agent's doing a good thing? But you know, if it's customer facing, what's the csat, what's the handle time? What's the metric that we're going to use for a key result? You know, is it retaining a customer, is it upselling a customer, but just defining a couple of those to go into it and then defining the use case, when you start to see that wheel turn, you now know that you're on the right path. So it's really that, that sort of life cycle of like, I understood, you know, I wanted to do this thing, I understood how I was going to measure it. Now I'm starting To see results. Now I'm continuing to refine, and I see more results, et cetera. That wheel spins. And when you get the wheel spinning, that's when you really start to get into creativity. And that's also when you can start to say, okay, great. How do I extend this to other agents as well?

So at Salesforce, you know, we hear a lot about all the AI hype, and as we talked about, we're a bit ahead of the curve because we actually have AI delivered within your business, within tools that you're already using. But I would imagine that also puts a tremendous amount of pressure on us to get it right. And I think you're at a very unique position because you're also, you know, guiding customer zero. So not only do you have to support us as an organization generally operating well and living our values, but also as you're pushing new agents into what we do here at Salesforce, the pressure to get it right is really big. Right for our customers and for how we showcase Salesforce on Salesforce. So how do you think about that and how do you go after agents for us as a customer and making sure that, you know, we're deploying it in the right way and finding success?

We went through a growth spurt where we had as many agents as possible, and so we had, you know, 100 plus agents, many of which were just not that useful. And I think that was part of our learning curve. The great thing about Salesforce is we do those learning curves very quickly. So we very quickly realized that wasn't the way to go after it. And so we focused on sort of five key agents for us internally. But I think in any technology, you know, people a lot of times like to talk about, you know, bits and bytes and the very hardcore technical stuff, and that's all critically important. But I actually think as a technologist, especially now, the most important quality that we need to have is empathy. And so one of the things that we have at customer zero is we've going through that same journey. We're going through it hopefully a little bit faster than our customers because we are leading not only our own efforts, but hopefully helping lead the industry a little bit. But that gives you a good deal of empathy for the customers coming behind you, so that when they have a conversation of, hey, we're struggling with this, and it's like, yeah, we did too, and here's how we got through it, I think is a critical component of customer zero is that empathy. And also, you know, sprinting ahead of the product in certain instances where we're going to go out there and we're going to try a couple of things and figure out if they work. Like, it's so new that the odds that we're going to be right on every single thing we ship is impossible. So we need to try a bunch of different things and figure out which ones work and then codify those into product offerings for our customers. And again, I think that's a type of empathy where it's like, rooted in, like, empirical trial and error, but it's also one of those things that develops a little bit of, call it road feel. We know how to drive this car and we can talk to our customers about that journey in a way that is relatable to them. Because it's that practitioner mindset.

Absolutely. What are some resources, maybe your top three resources, that you might recommend for someone who is approaching learning more about AI, becoming more of an expert, and feeling more confident in the way that they can speak about it? What would you recommend?

Look, I think a number one is play around with it. You know, we do provide a lot of tools inside the company. You know, you can use Gemini. Everybody's got access to Gemini. Hopefully they're using it for, you know, things like meeting summaries and all that sort of stuff. Play around with Agent Force, which has, you know, multiple LLMs sitting behind it. I use ChatGPT a whole lot for cooking, especially fantastic for recipes that you need substitution. So somebody's gluten free. Hey, give me this recipe. Gluten free. But just get out there and practice. You know, I think a lot of times, especially, you know, as people tend to get older, we get a little scared of failure, where when you're young, you don't even think about failure, you just go do. And so I think we have to get ourselves back into, you know, like, like Mark says, the beginner's mind, part of the beginner's mind is not being terrified of failure. So just go play with it. And yeah, you're gonna make some mistakes. And yeah, some of this stuff's going to be weird and not make sense to you, but eventually you're going to know how to do it. You know, again, the analogy of driving a car, when you first got behind the wheel and you were driving a car, you were terrified, you know, white knuckling it. Oh, my God, what do I have to do now? Everything's slow hesitating, second guessing. Now it's like you're driving a car, you're drinking a latte you're on the phone, don't ever do this, but you're multitasking. You don't even think about the driving the car. And so the. A number one thing is just get out there and do it. I think number two, you know, Trailhead is a great resource. The agentic side of it. Steve Fisher, my boss, has been pushing this. You know, take the agent certification side of it. There's a lot of great information there that I think is really fantastic. And the third thing is it is in the zeitgeist right now. So find people that you resonate with that are speaking to you, whether it's a podcast or a blog or just a, you know, Sam Altman or somebody that, like, you feel resonates and speaking to you in the way that you want to be spoken to, and just dive in, immerse yourself in it. And that. That goes back to C Point 1, which is just play with it. When you see one of these things, don't just listen to it. Go do it. Go try it out. And it's amazing. I use ChatGPT. I took a picture of my washing machine when I got here at Cannes.

I did the same thing.

Yeah. Because none of the symbols make any sense to an American. And it was able to tell me how to wash my clothes.

I was thinking about that this morning as well. I'm like, when have we before ever been able to take a photo? And I'm like, give me the steps, 1, 2, 3 on what I need to do in order to wash those clothing. And it was incredible. So you heard it here first. Start using ChatGPT for recipes. I use it to plan a trip to Disneyland. And honestly, I cut a lot of lines with that. And so I think this is a great note to end on just how incredible it is for you to put yourself out there. Start getting familiar, use it in easy, low lift, simple ways that make you feel comfortable, like a new driver. And then before you know it, you'll be off to the races. Joe, thank you so much for joining us today on our little Tiny Mike show.

Thanks for having me.