Welcome to the AI Enterprise | Salesforce Connections 2024 Keynote

When it comes to customers, there can be nothing artificial about the experience.

It's absolutely critical for us to be able to connect the dots. From a marketing CRM perspective, data allows.

You to understand the needs of your customers.

Our customers are saying, wow, you know, they actually understand who I am and what I want.

And so with CRM, AI, data and.

Trust, we can create an exceptionally personalized experience for our customers.

We had a lot of our customer data and information in different places. Bringing all of this together will take the power of Data Cloud.

We had over 30 different systems that had customer info and order info using.

AI and Einstein has given us the.

Ability to really take on that single source of the truth.

I needed a trusted source to secure my data and Data Cloud was the perfect product to take that journey with me.

Dare Cloud becomes that central hub where all of that data flows in. And it helps create this dimensional view of our donors.

With AI, with data, with the latest innovations, Marketing Cloud is really the engine that powers the customer experiences we want to deliver. So this is Marketing Cloud. We're able to look at customers who perhaps have gotten interested in fashion and we're able to then tailor the push messages all to do with fashion.

The insights enables us to build specific targeted campaigns.

Thanks to Salesforce Commerce, we're able to deliver real time information automation capability to make our customer happy.

Salesforce einstein1 commerce is the backbone of our purchase functionalities.

It is all about having really this.

One landscape of data to derive a.

Customer 360crm with AI and data powered by trust, that's the way forward for us.

Please welcome president and cmo salesforce arielle kelman.

Hello. Hello Chicago. Welcome everyone to Connections 2024. This is our annual conference for marketing and commerce professionals. I am thrilled to be here today. We got a lot of information to cover, a lot of content. Just to kind of ground you on everything the subject to today's presentation. Welcome to the AI enterprise. We're going to talk about how AI is changing the way that we work and how you can harness this technology. But first I want to say thank you. Thank you to our customers, our partners, our trailblazers. Thank you to our MVPs. Do we have MVPs here? These are the secret weapons of Salesforce deployments. If you don't have one, I encourage you find one or produce one. Producing new MVPs all the time. Okay. Since we started the company in 1999, we've been focused on becoming the number one CRM provider. And we've grown over the years. And this year we're projected to reach $38 billion. And we're proud of that. But we're also proud of the innovation that we've delivered to customers and for being recognized as one of the leading philanthropic and ethical companies in the world. And a lot of this is because we're really grounded in a set of core values. This is a value driven company and always has been. And at the heart of that is trust. You trust us to make you successful, you trust us to help you make your customers successful. And now more than ever, you trust us with your data. And we're going to talk a lot about that, because in the world of AI, data is absolutely important. It's the key to success with AI, but it needs to be done in a trusted way. You also trust us to innovate and to enhance our products and to build our products in a way that is both equitable and sustainable. So philanthropy has really been in the heart of this company's DNA for a long time. As many of you know, we created this one, one, one model that's been replicated by thousands of companies around the world, especially startups where when we started the company, we put 1% of our 1% of our employees time and 1% of our product into a foundation. And the results have really been amazing. And I think all of you participated in these programs. We've donated over $700 million. Our employees have volunteered over 9 million hours of their time. And we have almost 60,000 nonprofits and educational institutions running for free on Salesforce. And one nonprofit I'd like to call out is Big Brothers and Big Sisters, one of my favorite nonprofits. Thank you for everything you do. Now they are actually in our campground today, so I encourage you to visit them and learn more about their amazing programs. When we started the company, the vision for the product was really to help companies connect with customers in a whole new way. I thought we'd bring up a customer early in the presentation to just ground us in how customers are really using our products. And so the company that we're going to bring up today is a company called Turtle Bay Resort. They have a resort in the north shore of Oahu. Wonderful place. And they, like many of us, struggle with challenges like disconnected data, trying to bridge different teams to be able to work together and deliver a great customer experience. And of course, in not very surprising news, they've turned to Salesforce to help them. So I'm really honored to have with us today Robert Marussi, the chief Commercial officer from Turtle Base. So Robert, why don't you come up here? Thank you, Robert.

Aloha.

Aloha. So why don't you tell us about, maybe we'll start with. Tell us a little bit about some of the problems and challenges you've had with your goal of delivering a really great customer experience to everyone who visits your resort.

Yeah, I mean early on we did quite a renovation and a re imagination in this resort and I had the luxury of being able to change my distribution channel, my messaging, my digital brand experience. And I knew early on that I had to change my tech stack. My tech stack was closed loop at best. Right. It was a lot of point to point solutions. We took off the shelf to accomplish these very siloed engagements. And it didn't speak to anything. There was no end to end connection. There was no view of the customer. And I knew early on that I needed to take the journey to get a 360 view of the customer.

Great. And you've been one of the earliest adopters of our data cloud product. And we're going to talk a little bit about Data cloud today. So can you tell us a little bit about how that plays into achieving this improved customer experience?

Yeah, I mean most of you know, right, we're in a choice economy, right. And it's, it's about the client, the consumer being in the middle of this Omnichannel spoke and it's our responsibility as brand people to be delivering choices to that consumer. And that's what they want, that's the way they want to consume their data across the omnichannel. And I knew that I had to get into a whole advanced personalization, hyper personalization tactic and strategy. And I knew that I had to be able to segment my data based on mosaics and Personas speaking to the client and what their interest and affinities were. And Data cloud became that very apparent platform for me to segment my data, put it in different data sets, data spaces, and then be able to talk to the consumer in that hyper personalized fashion.

Excellent.

Yeah.

Any advice for our friends here in the audience on how they can achieve some of the same results?

How much time do we have?

A couple hours.

Really? At the end of the day, it's how do you keep your use case simple? Because it's a daunting thing to think about AI. And what is AI in terms of how you're digesting it? Right? Is it gen AI, is it AI that lives on a trusted data model? What are you trying to use it for? And in the MarTech MarCom space. It was very easy for me to get there. So I kept it simple. I kept the use cases very tight and simple to be able to really talk across platform of marketing and all that. The advice is keep it simple, really understand, get the reps in the game, spend the time on the insight sites, McKinsey Insights, Gartner Insights, Accenture Insights, and learn about data and AI. Because the conversation with the C suite level leaders today is long and it takes a while. So you're going to be in an education process, you're going to be in a change and transformation engagement with him and really try and understand it and learn it every day. It was really the biggest takeaway for me to be able to talk, to talk to that level of leader and be able to digest it down in chunks so they understood the ROI and what was happening.

Makes sense. Start off, start off small. Maybe pick a couple data sources and a couple of ways you want to activate and implement that first and learn and grow.

Absolutely. And get the wins along the way. You get immediate wins along the way. Right. And that's the key. Right. How are we going to make money? That's the question you always get as operators. But if you can prove that success quickly, then, then you're off to the races. So.

Excellent.

Great.

Well, thank you for spending time with us.

Thank you.

Appreciate it. Okay, well, let's, let's talk about AI. So AI really has an opportunity to fundamentally transform how we work. And we've been going through various waves of a really predictive AI was the first wave and we launched Einstein in 2016. And I think it's become fairly commonplace now to do things with predictive eye, like scoring leads, like coming up with next best action for our salespeople, like routing cases. This is something that's become very commonplace and I'd say mainstream. But now we're in another wave. We're in a generative AI wave where the AI is actually talking to our employees and where we have the opportunity to be able to augment the capabilities of our employees so they can be more productive and more effective. And then after that, the next wave that we're just starting to enter is autonomous. And this is where the AI can take action on its own. The AI can do work for us, talk to customers. And as we head towards the fourth wave of artificial general intelligence, we got to remember that at the heart of this, it's all about effective, lasting customer relationships. So let's talk about how to get there. So I want to go to Consumer AI first, because I think we can learn a lot from what it can offer, but also what it can't when it comes to how we would use it in the enterprise. So I'm sure all of us here have used consumer AI tools like ChatGPT and they feature three parts. There's the user experience that you interact with. There's the model, the large language model, the LLM, and there's the data, and there's a lot of data. And what's amazing now is we have tons of choice. There's many options at the UI level, there's many models and all of these have been trained on a massive amount of data. Blog posts, videos, open source databases. And this is great. So for example, if you might be having a minor kitchen pantry disaster of needing to make chocolate chip cookies and not having any eggs or baking powder, generative AI in these consumer models, they'll do a fantastic job of helping you figure out how to go make these cookies without those key ingredients. But when we go to the enterprise world, all that data really isn't that useful. It doesn't help you, it doesn't help the AI give you the right answers, give you the right advice, produce the right results. When you want to do things like deliver a better customer experience, when you want to have your employees be more effective at solving customer problems, you need your own data. You need data about your customers and what's unique about them to ground the AI to get the right results. But you got to do that in the right way. We're going to be talking about that. So in addition to being able to trust the data, we also have to get it all in one place. And all of us, I think, are struggling with islands of trapped data across the enterprise. So we think about the core. A lot of us have our core customer data in Salesforce, but a lot of us also have tons of customer data in different places. In our data warehouses, in Snowflake, in our data lakes, in places like databricks and Bigquery, in our back end operational systems. There's data everywhere. And so this is a challenge. So if we're going to be able to have the ability to ground AI and trusted data, we need to be able to have access to all of it. So in addition to being able to get all the data together to unify it and to deal with the siloed data problem, there's also a lot of other questions that we need to answer. To have AI be effective for the enterprise, we have to make sure it has the Right security and privacy. We've all spent a ton of time setting up the right security and data access rules in Salesforce. So the right information goes to the right people. We don't want to undo all that by implementing in the wrong way. We also don't want toxic results. We don't, you know, the worst way to make a customer relationship is to offend the customer. We do not want to do that. So no toxicity and no hallucinations. We've seen examples where AI has made offers to customers on websites that don't even exist and the courts have shown we're really liable for fulfilling on those offers. So you got to have the right data, it's got to be connected to CRM. So all of these concerns need to be taken into account in order to deliver AI for the enterprise that's trusted and really works and really delivers business results. So this is why we created Einstein One. So Einstein One is the trusted platform for the AI enterprise. And what it is, is we've taken all of our applications, all of our AI capabilities, and with Data Cloud, we're bringing together all of the data that you have in the enterprise, the data in Salesforce, and also reaching out to data outside of Salesforce. If you've done the work to organize your data in a data warehouse like Snowflake or Redshift, great, we can reach out to that data and it can reside in those systems, but you can use it in Salesforce just like it was the standard object and do all the things that we'll show you a little bit more. And then most importantly, this is all built on a metadata framework because the AI needs to understand about your data. They need to know things like accounts or companies. Contacts are people that work in those companies. The cases are problems that those humans that are contacts have and that kind of intelligence needs to be built into the platform and we've built that in. So not only are we giving you intelligent, truly useful AI because it's grounded in your data, it's automated so it can take action, you can implement quickly because we have low code and no code for customization and deployment and making changes fast, but also it's open. So we built this architecture to work with any large language model. If you've trained your own custom models, we can plug it into those. If you have a third party provider you want to work with, you plug in those. And we have large language models of our own. But being open is an absolutely key part of this because we know that you have existing investments. So let's talk a little bit more about Data Cloud. So Data Cloud really is the heart of modern marketing. And as a marketer, I have a Martech team and they use Data Cloud and it's really become the foundation of how we do marketing at Salesforce. So we think about having all these sources of data. Some of it's in Salesforce, some of it's outside of Salesforce in different system purchase data, CRM data, inventory data, data from our website, data from our snowflake data warehouses. And we pull all this together and once it's unified with Data Cloud, you can do everything with that data that you can do with any standard object. And so the main three things that we're doing is providing visibility, activating on that data and grounding AI in that data. And I want to walk you through each of them, but they're all really important. So let's take visibility first. So we have a wide variety of data. We have over 80 different data sources that my Martech team has unified into Data Cloud. And then in terms of visibility, well, we want to have access to that data for our sales and service people. So for example, if you click on the account screen, wouldn't it be useful to our salespeople to see a summary of all the engagement that this customer has done with our marketing programs? Wouldn't it be great if they could see the purchase history, if they could see the recommended next best action from a custom deep learning model that we've built that offers up to our salespeople some guidance so we can bring that all together in a very easy way. So next is taking action. So one of the things, and this is sort of the classic customer data platform use case and Data Cloud is great at it. So if, for example, I want to market one of my favorite products, Sandbox to our customers, I can start with a machine learning built segment in Data Cloud. So Data Cloud will help you create the segments and then I can activate it in a wide variety of ways. We have a lot of connectors. One of them, one of my favorites is LinkedIn because LinkedIn gives you a platform where you can target specific companies and the people you know and don't know in those companies that have certain roles. And so with this Data Cloud connector, we're able to very seamlessly run ads on LinkedIn to the specific humans in that campaign, sorry, in that segment that are a great fit for this product, offer them up Data Cloud and then they can go and make a purchase in the your account area of our product. So we talked about visibility, we talked about activation, but there's also AI. And to give you a sense of why grounding AI and trusted data is so important, I thought I'd kind of give an example again, kind of go back to the consumer versus business use case. So let's say that we wanted to use generative AI to produce a marketing email. Now if we took a consumer large language model and we just asked it a question of write me an email inviting our client Lauren to attend an upcoming conference, it's going to give us pretty basic, pretty boring, pretty generic email that's just not that compelling. But if instead we put in the prompt information about the customer, that's useful for writing a good email. Just like what we'd think about if we were to write the email ourself. So what type of company are they at? What is their role? What products have they bought before? What products have they been looking at on our website? And what do we want to invite them to? If we take all of that, we can get a much more useful, I mean, it's very like kind of to go to plain language sometimes, a much more useful email that's likely to be compelling and get results. So not only have we been delivering a lot of AI functionality, but we've also been innovating across our product with thousands of new capabilities. And I'll call out some of the marketing and commerce capabilities that we're going to be spending time on in this conference. So for Data Cloud, We've delivered over 200 new connectors to make it easier than ever for you to integrate Data Cloud with all the data sources that you have. We've launched today our Einstein Copilot for our commerce products, giving a copilot for merchants, for buyers, for shoppers. And then we've also delivered several new enhancements for our composable storefronts functionality in our commerce products. So how do you build an AI enterprise? Well, we've broken it down to five easy steps. Build a customer 360 to give you a unified view of your customers. Unlock and activate all of your data so you can have unified data for your activations and for grounding AI. And to use that AI, you need a copilot. But not just a generic copilot that doesn't know anything about your business. A copilot that's deeply integrated to your data, to your metadata, that's right in the flow of work to augment the capabilities of your employees so they could be more productive, you need an AI that's built into and helpful for your analytics to help you find the right insights, what questions to ask. And then lastly, AI should be part of collaboration. One of the most exciting things we're going to show is Slack AI and the capabilities are really outstanding. I use it every day. I think you're going to enjoy that. Let's start with the customer 360. So at Atara, the customer 360 is really about giving you the capability to interact effectively with your customers as they touch all of your different channels and all of the different employees. And we've been innovating at each of those layers. So really it starts with sales cloud. And with our AI sales cloud, we give you a copilot right in the flow of work that understands what you're doing that could help you build and close pipeline faster than ever and boost productivity. So one of the companies that boosted productivity was Spotify. And by integrating their workflows for sales directly into Slack, they were able to increase productivity by 40%. And on the service side, I think one of the most obvious and fastest adopting areas for generative AI is customer service. Because in a customer service context we can again have that copilot supercharging the capabilities of call center agents to be able to help customers solve their problems. And then also we can have the AI talk to customers directly through Einstein chatbots and solve their cases quickly and deflect those cases so the agents can spend more time on higher touch customer engagements. And the example of that is Sonos, where they were able to use Einstein chatbots to deflect 40% of their cases to self service, which was a big win for that company. In an AI marketing cloud, our generative AI capabilities are helping marketers create segments and activate on that data and build campaigns, customize email templates, a lot of content oriented and segmentation automation. And one of the customers that took advantage of that is St. Jude's Research Hospital. And so by creating highly personalized campaigns they were able to raise $2 billion in the past year, which is truly amazing. I actually wasn't sure about that. I looked it up and that is true. It's truly a phenomenal fundraising effort. Then for commerce, as we're going to show you later, we've added Einstein Copilot to our commerce capabilities and what it's going to do is it's helping merchants, helping commerce professionals build their stores, build their detail pages and also copilots to interact with customers so we can help convert those shopping carts to revenue faster than ever. So we're going to go into each of these parts. The next part of this is Data Cloud. We're going to show you more about how data cloud can be the heart of modern marketing commerce. With that, I'd like to turn it over to Kelly Eliyahu from our Marketing Cloud product marketing team. Thank you, Kelly.

Thanks, Ariel. So businesses have been trying to connect Data for over 30 years. Back in the 90s, data was managed in on prem databases, but the problem was there was no interoperability. Then came DMPs, and these were great for activating anonymous customer profiles for ads. But the problem was they couldn't handle any type of known data. Then came CDPs and these were a real breakthrough because they helped us connect both known and unknown customer data for marketing use cases. But in today's world, that isn't enough. We're missing opportunities. By not connecting CRM and cdp, we're following customers around with ads for something they already bought. Or we're making them repeat the conversation with every single handoff. Now, I'm a marketer and I know that in order to create those truly personalized and connected experiences, we need more than just marketing to have that unified view of the customer to be able to action on that data. We need marketing, but also sales, service and commerce as well to have that data. And this is exactly why we built Data Cloud. Data Cloud is more than just a customer data platform. It's a company data platform. It's a hyperscale data engine built directly into Salesforce that makes every cloud better. It helps you unlock data from any source and activate that across any application. You can do things like pull in inventory data to help inform your campaigns, or bring in web data into CRM so that sales knows exactly when a prospect is looking at a pricing page and can take action. And we're so excited to announce that we now have hundreds of new out of the box connectors available for Data Cloud. Now these are built directly natively into the platform, so there's no need for any custom integrations. Okay, so how does all this work? Data Cloud connects data through connectors, and these cover everything from Salesforce apps to third party data lakes and warehouses, all without any data duping. This includes unstructured data as well. So think an email or a PDF knowledge article, and then all of that data is harmonized into one metadata model and unified customer profiles. This can then be activated in CRM to power both actions and insights. And this is something Salesforce does better than anyone in the industry. And that's because data Cloud is built on Salesforce's Einstein One platform and integrated with our metadata model. Now, metadata may sound like an intimidating word, but it's really not. It's quite simple. Metadata is just context about data. It's what gives it meaning. If I were to show you five random numbers, you wouldn't know without the right context, if I was talking about a price or something else. Computers are the same. They don't know unless we tell them. We need metadata to tell us that those five random numbers represent a zip code. Metadata translates that raw data into contextual information that used anywhere inside of Salesforce. So that can be a lightning component for displaying our data within apps, or that can be flow for powering workflows and automations. And all of this trusted data and metadata is what makes AI effective in the business context. So if I was to ask ChatGPT about I don't know my campaign performance, I probably wouldn't get a very good answer. And that's because consumer data, or consumer AI, we hope, at least, doesn't know anything about our campaign performance. Right? It's learning from data on the Internet. Data Cloud allows you to safely insert important business data and context into every prompt you create. And then once that prompt has been augmented, we send that over to the LLMs. But you don't need to worry. None of that data is stolen or retained by the model. We ensure it remains your data. Okay, so we've talked about why we built Data Cloud, how it works and how it makes AI more effective in the business context. Now I want to talk about what you all actually came here for, which is announcements. So if you haven't been picking up what I've been dropping down, it's that data cloud makes marketing and commerce better. And we've got two new innovations to put our money where our mouth is. First, we've got Data Cloud for Commerce. This powers every part of the commerce experience with trusted data. It powers segments, it powers insights, it powers journeys. And second, we've got Einstein Personalization. This is built into Data Cloud to help you build one to one customer relationships through AI decisioning and real time recommendations. And both of these innovations are available this summer. And the innovation doesn't stop there. We also have our new zero copy partner network. Now this is made up of partnerships from major data platforms like Snowflake and like Databricks, but also implementation ready partners as well as third party partners on our app exchange with AI ready data sets. But we don't just want you to take our word for all of this, we want you to get started fast and see value fast. And we want to bring this to life through a customer story. So Aston Martin is an iconic automotive brand known for combining precision, technology and craftsmanship. And as you can imagine, personalization and quality really matter to Aston Martin. And that's why they turned to the Einstein One platform to help them unify all of that disconnected data to power ultra personalized customer experiences. So what I want to do is bring this to life through a short demo. So can you please help me give a warm welcome to our amazing demo team? They are going to be driving. They'll be driving live demos throughout this entire keynote. Okay, so what are we looking at here? Here is Aston Martin's view of a customer inside of Salesforce. And we're looking at Abigail's profile here. So if we look in the top left, we've got some of the basic information that you would expect. Right. We know she's from Chicago. We've got her email, we've got her phone number. And below that, we have a little bit more information as well. Right. We've got her past appointments, and we know that she purchased her last aston Martin in 2021. But if you kind of look across the rest of the profile, there's not really that much there. This definitely is not a complete picture of everything Abigail is doing, and it's not enough to create that ultra luxury experience that Aston Martin cares about. To do that, we're going to need more data. But I don't want to just ask our demo team to go and manually pull that for me. I think we can do better than that. So let's go ahead and hop into data cloud. All right, so what we're looking at here is data streams that are pulling in data into data cloud. This isn't very big list. We've just got our 1Salesforce.org. So what I want to do is add more data to help us understand things like, how is our customer using their car? Do they need service? How can we enhance their experience? So let's click new and add more data sources. So here we've got a whole bunch of data sources. We've got our Salesforce apps, we've got third party data lakes, and if we keep scrolling down, you'll see we have hundreds of connectors available through mulesoft. But what I want to do is add databricks. Now, this is where Aston Martin is collecting all of their telemetry data coming from their connected cars. And by bringing this data into Salesforce, Aston Martin is able to increase the ROI on their investment in databricks. That's because all of that previously trapped data will be available to help them create that ultra luxury experience we talked about. So let's go ahead, click Next and add databricks. So remember that idea of harmonization I talked about earlier? Well, this is that process. We're connecting the data in databricks to the data model inside of Salesforce. Basically, this is to help us make sure that the right telemetry data is associated to the right customer. All right, so this mapping looks great. Let's go ahead and press Save now. While we were busy harmonizing all of that data, our demo team was rapidly adding new data streams. They're the fastest in the business. So in addition to databricks and that one Salesforce. Org, we've got a whole bunch of other data here. Right. We've got our other Salesforce orgs as well as APIs. But what I want to do is actually go back into that customer view and and hit refresh because I've got a lot more data to work with here now. Wow. This is so much more of a complete view of our customer. And this is totally unique to Salesforce. So let's check it out. In addition to that standard contact information we have below that, we also have calculated insights. And this is just a fancy way of saying that we're using the data we just brought in to help us understand things like C SCAT score or customer lifetime value. We've got our previous service interactions and then in the middle here, we've got that telemetry data that we just pulled in from databricks so we can see current usage of the car. Then we have our engagement activity feed. This allows us to see things that Abigail is doing. It looks like she might be showing some interest in the new DB12. So again, much more complete view of Abigail. But the real value of data is the ability to action on it. And this is something Salesforce does better than anyone. So let's go ahead and take a look at what that view looks like in flow. This is our workflow automation tool and it helps us do things like transform interactions into journeys. And what we've got here is a data cloud triggered flow. This triggers anytime a customer takes a specific action. In this case, it's Abigail showing interest in the DB12 and we're going to show her an invite to the local DB12 launch event. So I want to jump over and see what that experience looks like for Abigail. So here Abigail is receiving a personalized WhatsApp message. This is her preferred communication channel and it's sent at the time she is most likely to engage. Because this is using Einstein personalization, we're able to use all the data that we know about her to make sure it's the perfect offer. In this case, an invite to that DB12 launch event at the Chicago dealership. And the personalization continues across the Commerce experience as well. So when Abigail goes to AstonMartin.com, she's immediately shown a DB12 data cloud for commerce is again is using all that information we know about her to make sure this is an ultra personalized shopping experience. The DB12 is in her favorite color. We've got that beautiful Chicago skyline, and all of the configuration is around, maximizing performance. And as Abigail configures and browses more, all of that data is captured back into Data Cloud and that's made available to the dealership. So when she walks into that showroom floor and it's a continued personalized experience, it's a connected journey. All right, so we want to get you all started with Data Cloud today. That is our goal and we've made this easy in a couple of different ways. So we've got our Data Cloud starter bundle. This is a bundle of professional services and licenses that'll help you get started with a common prototype use case of Data Cloud. And we've also got trailhead. We've got 10 plus hours of courses, of videos, of hands on workshops, all to help get you started today. So I want to see all the phones out, get your phones out, take a picture of this QR code and get started with that. Ariel, back to you.

Thank you, Kelly. And one more call as we have Hands On Labs Experience Experience is here. So if you want to have someone walk you through everything from a five minute overview of getting started with Data Cloud to a deeper exercise, it's all available here. We have space for 4,000 people, so we have a lot. Well, next up, we're going to talk deeper. We gave you a deep dive on unifying your data and taking action with Data Cloud. Now we're going to talk about AI and show you AI in action with Einstein Copilot. And it's my pleasure to invite up Michael Affronti, SVP and GM of Salesforce Commerce.

Thanks, man. Thank you. Appreciate it. Morning. How are you guys doing? All right. This is my favorite part of this keynote, not because I'm talking, but because we're going to get really geeky. Are you guys ready for that. All right, let's do it. Okay. So as commerce and marketing leaders, we know that our goal is to deliver deeply personal relationships. Right. We want to make every customer interaction feel like it's the first time that we've ever spoken to them. But we know everything about them. But we know that delivering one to one personalization at scale is really costly and difficult. Right. But customers are expecting us to meet them where they want to be. And that's the real important part of everything we're talking about today. What I'm really excited to show you is how at Salesforce, we're going to help you deploy data and AI to actually deliver that one to one personalized experience. And the way that we're going to do that is with our new copilot experiences. So our Einstein copilot is one digital assistant, a conversational digital assistant that works across all of your trusted company data. The coolest part about copilot is that you can literally ask it questions and it will give you answers. It can translate content, it can write summaries, it can even action salesforce records for you. And the thing we've done is built a single copilot experience that works across every salesforce application, which means you can get started today and actually learn the same behaviors across all of our applications. Okay. And what's really exciting, and this is perhaps my favorite part of what I get to talk to you about, is that we are delivering new copilot experiences today. For my two favorite, clouds, marketing and commerce cloud. Okay, so for our marketing cloud customers, our new copilot for marketers is like your digital campaign assistant. It helps you write campaign briefs, campaigns, it'll generate content for you, generate images, it'll even write entire emails to your customers. Now on the commerce cloud side, we decided to give you two different copilots. Now, has anyone ever walked into a store and had this incredible experience with an in store associate where they sort of know everything that you need, they can help you find specific products, and they also remember everything about the last time you were there? Generally, yeah, we've all had those experiences. Our copilot for shoppers is exactly that. It is a generative AI assistant that sits on top of your commerce cloud storefront and actually talks to your customers like a 24. 7 in store associate. It's truly incredible. We've been building this in pilot with a handful of customers over the last 12 months and the results have been phenomenal. Now on the merchant side, we've built a copilot for merchants and this helps you do everything from create new storefronts to write product descriptions, and even optimize your SEO with AI. And we'll show you some of that today. Now, the really great part about these Copilots is that they fit into a family of Copilot experiences that work across the whole Salesforce platform. And this includes our service cloud, our sales cloud, even Tableau, and our developer platform. So as you learn more about Copilot and start to get more familiar with it, it's ready and waiting across all of our applications. And it's a really special part about the experience. Now, there's another important part about how you talk to Copilot, which is using natural language. So when you ask Copilot a question, it goes behind the scenes and builds a plan. Now, a plan consists of your organizational data. It consists of your records, your workflows. This is how it's going to answer your question or your request. And that action that it can take for you are things like sending an SMS to a customer. It can update Salesforce records. It can even write summaries and send out emails. It's really powerful. But we also know that not every company and line of business is the same. So you can easily customize this whole experience without writing a single line of code. And we're going to show you that in just a few minutes. But I want to talk to you about something really important to me, and that's the safety of all of your data. You know, at Salesforce, we've been protecting your Data for over 25 years, and that hasn't changed a bit with AI. So Ariel and Kelly talked a bit about this, and I want to open up our trust layer a little bit for you guys. So we have built what we call the Einstein Trust layer. This is a single platform that sits underneath every Copilot experience you're going to have. And its main goal is to protect your customer data from the second you enter a prompt. Our data masking and secure data retrieval ensure that none of your raw customer information is ever passed outside your organizational boundary. And our dynamic grounding actually helps reduce the risk of hallucinations because we are grounding it on your actual company data. Your existing CRM records, existing merchandise data, existing marketing data. Now, this one's really important. Zero retention policies with all of our third party LLMs mean that once the response is generated from the LLM, they discard everything that we sent to them. So they never retain or train their models on your obfuscated customer data. And when that response comes back. It is all passed through our built in toxicity and bias detection. And this part is really important because it gives me the confidence to help customers deploy and use our new Copilot for shoppers, which is literally talking to their customers so you can be safe and confident that we have got your back when it comes to the entire loop of safety around AI. Okay. I think it would be cool to do something with Copilot together. Like I said, we're going to get geeky. You guys ready? Yeah. Okay. Chris, Miles, you guys ready?

Okay.

How about a hand for Chris and Miles at the demo booth? Okay. We're going to pretend that we are marketers at Aston Martin and our goal is to build a campaign today. And I want this campaign to help our customers be really excited to during what Aston Martin calls the lull. And the lull is the period of time from when a customer ordered a car while it's getting built to when it gets delivered. Okay. So let's ask Copilot for marketers to help us out. Create me a campaign for customers who recently ordered the DB12 car. What copilot's going to do is using our customer data grounded on our brand and information. It's going to go and create a campaign brief. I like to think that this is sort of the pitch from an agency. Right. Here's what we think we're going to do. This seems like a pretty good campaign. I'm going to give it a preview and what we see underneath this is a summary of what this campaign is going to do. It's going to create this segment for me. It's going to write an email subject line and even the preview of a paragraph that it'll send to the customer. I think this looks pretty good. Let's go ahead and create the campaign. Now this is where it gets really cool because we're going to go in and edit this built in email template that it generated for us. And what we're going to show is that this email template is of course rich in the Aston Martin brand and tone, but it's also got some really cool data inside of it. So you remember Kelly talked about the hundreds of data connectors that we have. One of the things Aston Martin has done is connected their vehicle assembly status data into data cloud. So now when this email goes out, it has a real time update on the actual status of her car across the assembly line. That's really cool. But I want to add one more thing. If you've ever done any commerce, merchandising or promotions, you know that during that this build time is a great opportunity to offer them some upsell, some cross sell. So we're going to add a recommended add on section using Copilot. I simply ask it to build this segment in the email and it goes and creates it automatically. Now this is pulling information from the commerce cloud product database, like the images, the metadata, and even the title. I think this looks good. The great opportunity to upsell some sunroofs. Let's go ahead and save this email. All right, so now if you've ever done commerce merchandising before, you know that one of the things you're always thinking about is how can I make these offers even more enticing, right? So one of the things that Copilot does for you on your behalf is it looks at your commerce data, it looks at your service cloud data, and it notices things. And in this case, it's actually going to tell me proactively that it thinks it found a way to make this campaign more efficient. And that's because a large percentage of customers have been going online and buying the carbon fiber tailpipe package in conjunction with the interior package. So that's a really great bundling opportunity. So instead of having to go into Commerce Cloud and actually set up the bundle and do all these different things, I'm simply going to tell Copilot to go ahead and activate that promotion. So now every customer that gets that promotion 10% off all the cotton farber that they want. Okay, this looks really good. I think we're going to go ahead and launch that. So now what I'd like to show you is what this looks like on the shopper side. So let's switch over to Abigail. And Abigail, as Kelly mentioned, has been experiencing the Aston Martin interactions over her favorite channel, WhatsApp. But what's really important is that this is running our Copilot for shoppers. So not only is she getting the build status updates over WhatsApp, it's also an opportunity for her to say, hey, is it too late for me to add the carbon fiber interior package? Copilot for Shoppers looks into the assembly system, notices that this is a great time to do it, and then says, absolutely. But more importantly, we built up promotion. And Copilot for Shoppers says, why don't you add both for that 10% off? Abigail says, yes, add both. And just like that, using her save payment methods, thanks to our new Salesforce checkout SDK, she's able to securely check out just by saying, please give Me both of those products. And that's it. That's how Copilot for shoppers and our Copilot for marketers, we created a marketing campaign, pitched it to one of our customers, and actually cross sold some products. So thank you very much. I hope you guys appreciated that demo. Okay, part number two. Now, this is where it gets, as I like to say, really fun and geeky. So buckle up. We're going to do some cool stuff here. I'm going to show you how we can customize everything that we just did for your individual line of business and your company using our Einstein One studio, which is a portfolio of tools that lets you do things like build your own prompts, create new copilot actions, and even design your own AI model. So I like this stuff so much that. Chris, is it cool if I do this myself? All right, Cool. All right, team. All right, so here we go. What I want you to imagine first is that I am a dealer at Aston Martin. Aston Martin sells cars exclusively through their 150 dealerships around the world. And one of the things their dealers do every day is they start in their Aston Martin portal, like any dealer. The first question that I'm going to go ask copilot is, who should I sell a car to today? Which customers should I reach out to that are most likely to buy soon? Now, copilot is going to think about this for a second, and it's going to give me the best answer it can, which is there are three people that have appointments today to come in and check out cars. And if you've ever bought a car, this may not be the greatest signal that I'm ready to buy one. It just may be that I'm coming in to browse or honestly, you want some free coffee? Who knows? Okay. But what we're going to do is now go teach copilot a new set of skills to help make this answer a lot more effective. We're going to come over here, and we're actually going to build a new AI model. Okay, you guys ready? Want to build a model with me? Yeah. Okay, here we go. All right, so we're going to go and create a model from scratch. And the model that we're going to create underneath the covers is what's called a predictive regression model. So what that means is we're going to take a bunch of historical data about Aston Martin's customers, put it at a model, and then have the model try to predict an outcome. And in this case, it's going to be how likely are people to want to buy a car? So I first have to tell the model, what data am I going to use? I'm going to pick all of the individual data that we created in that unified profile in data cloud. Then I'm going to hit next. You have an opportunity, depending on how big your data is, to filter down. But in this case, 516 records, we're going to go with all of them. Now this is where you tell the model, what are we trying to do? And here we're actually trying to predict the likelihood to buy. So we pick that field as the target. And then if you've ever built a regression model, you know, you either need to maximize or minimize. And we're going to maximize to one, which simply means optimize the outcome to predict people that are going to buy. Value equals one. So we hit next. Okay. Now this is where it gets even cooler, because the model can be trained on data using data cloud from all over the Aston Martin ecosystem. Average ownership length coming from the CRM database, car mileage coming from the telematics connected in a data cloud, a recent life event. If they've filled out a survey or a form telling people that they've moved or that they've maybe had children and are looking at an suv. All of these are interesting variables. So I'm going to add all of them to my model, then I'm going to hit next. Einstein can also pick the best model for you, so we'll let it stick with a glm. We're going to check all of the information here. Great. And then we're going to hit save. So model training can take anywhere from a few minutes to an hour or so. So I'm going to take one out of the oven that we put in just a few minutes earlier and show you the outcome. So model builder tells me that great job, guys. We built a highly predictive model, a 91% chance that we were going to predict a high likelihood to buy for an Aston Martin customer. This is great, but let's go see why. It actually shows you the top predictors. And these are the values in the training model that prove to be the highest propensity of influencing that decision to buy. And this makes intuitive sense, right? If someone's had a car for a really long time, has very few open service cases, and maybe had children or moved, those are all things that I think would make someone potentially want to buy a new car. So it feels good to me. The data shows that it is a high prediction so we're going to go ahead and we're going to activate. Okay, so now that we've built a new AI model, we're actually going to do another very cool and very geeky thing, which is actually build a new prompt template. So I like to describe prompt templates to our customers as mail merge for AI. It lets you build a new prompt which can be asked in copilot anywhere across Aston Martin. And what I want this prompt template to do is help people find their the best customer to reach out to and send an email to them. So I'm going to start by pasting in a command here. Generate an email that a dealer can send to a customer, inviting them to come test drive our newest offerings. This message should be clear and concise, but as Kelly said, it's really important for Aston Martin's brand and tone to be aligned across every customer outreach. So I'm literally going to write out Aston Martin's brand and tone and ask the prompt to always use it. Third piece, we need to point the prompt at a type of customer record to go and actually send a communication to. So we're going to go pull a unified profile and we're going to pull it out of data cloud. So this is the data cloud real time personalization model. It's actually a data graph that shows you where all the composite information is coming from. The record obviously has been pre built and looks really good to me. But I want to add one more thing here, which is the engagement history. Okay, so engagement history is how frequently and how recently has that person touched any outbound email or SMS marketing we've done to them. This looks pretty good, but let me just preview it. And when you preview this type of model here, you're actually going to see the raw JSON output of the application object graph. Okay, so this is just giving me one more triple check that it's going to output great information about the customer. So we'll save and deploy that. Okay, so now we're going to go back to the prompt template workspace. So remember, we built the prompt template, we're grounding it in Aston Martin's tone and voice. And now I'm going to ask the prompt every time that it runs to pull the unified profile of the record. Okay, so now we're going to pull in that customer information. Let's go preview it. So you guys see exactly what I mean When I hit Maria Costas on the left hand side here, zoom in for you guys. You're going to see the response the Raw response from the prompt. So this is the actual JSON coming back from data cloud that shows all the unified customer information that will be put into the generative response. And on the right hand side, you actually see the email that we would be sending to Maria if someone was invoking this prompt. It's actually pulling in dynamic car information from the marketing campaign. It's offering her to come by for a test drive. And it feels very Aston Martin. It's very on brand and it even includes my personalized. Thank you. So this looks really good. So we're going to go ahead and save that prompt template. Okay, so now Michael becomes a car dealer again and I'm back in my dealer portal. We're going to go ask Copilot the same question we asked it just a few minutes ago. Check this out. Which customers should I reach out to today that are most likely to buy soon? And Einstein Copilot is a lot smarter. It now tells me that I should reach out to Ashley Harrison. She has a very high likelihood to buy. It's explaining her lifetime value key kind of thing. In my mind, I want to have her at the ready and most importantly, that she interacted with one of our campaigns about two hours ago. So this feels like a really rich opportunity for me to go and reach out to and I actually want to do that right now. So let's. So let's email Ashley. Draft an email inviting Ashley to come explore our latest offerings. So now that prompt template is getting invoked. It's actually generating that highly personalized email and all the dealer had to do was ask. All the backend processing happens to make sure it's grounded in Aston Martin's brand and tone. This looks really personalized. It's nice and casual. The way I like to write. It includes my name as a follow up. So I think this looks really good. And we're going to hit this send and let's go see if Ashley's going to buy a car. Hopefully she will. All right, well, look, this is one of my favorite experiences because it shows how without writing a single line of code, you can completely transform our Copilot experiences to make them your own. You can use our prompt templates to give every dealer, every person at your company an easy way to pull unified information and use it in rich outreach. And you can use our model builder to build your own or bring models from outside your network to do an incredible amount of personalization that is only possible on our platform. All right, so with that, I'm going to turn it back To Ariel to take you guys to the next chapter. Thank you very much.

Thank you, Michael. That was fantastic. And just to remind everyone, everything that Michael showed there is available today. Prompt Builder, Einstein Copilot model builder is available for you to use today. Now, the next part of the presentation, we're going to talk about how AI can make your analytics and your collaboration even better. And with that, I'd like to bring up Kelly Thacker, who is in our demo booth. We're going to have another live demo and Kelly is the CMO of Salesforce Commerce Cloud. So take it away, Kelly.

Thank you. All right, you guys. So let's talk about the AI revolution. Now, with all of the incredible AI innovation that's happening, there is an increased demand from leaders, really everyone in the organization, to have better access to real time, accurate and trusted data insights. Now, you see there's this infinite loop that exists between data and AI, right? So the quality of your data is going to inform the quality of your AI insights, and those very same insights are going to inform how you manage and how you use your data moving forward. And at the end of the day, all we really want is data that's accessible and data that's usable. So when we're talking about data insights, there is no better product on the planet than Tableau. Tableau truly transforms how everyone in the workforce can solve problems using data. Truly becoming data experts. Now, the Tableau team has been working very hard on two new incredible AI innovations that I'm so excited to tell you about briefly and then show you. So the first is Copilot for Tableau. So you've heard a lot about Copilot today from Ariel, from Michael, from Kelly. So the very same incredible AI assistant is now available for every single Tableau user. This means you can streamline all your data prep to data visualizations, truly helping you accelerate data insights. You can talk to your copilot, you can talk to your CRM the exact same way that you talk to your analyst. The next innovation is my favorite. Out of all the tableau innovations they've ever deployed, this one is it because this allows you to truly interact with your data in a completely different way. Now, I get Tableau polls every single Monday morning at 7am and it delivers all of the insights that are most important to me as a marketing leader. So I get to look at my MQL's, my pipeline, my revenue, and it's going to tell me what's trending up, what's trending down, and any anomalies in My business. So now I don't need to go into a separate data visualization to try to connect the dots myself. Tableau Pulse is doing that for me. So I love it so much that I am going to walk you guys through an incredible demo of Tableau Pulse. So again, this is Aston Martin. This is how they use Tableau Pulse. At the top, you can see all of the metrics that have changed week over week. So right here, you see that the APAC segment is really surging on impressions. If you scroll down, you see all the metrics that matter a lot, that matter the most to the marketing leader at Aston Martin. So I'm going to want to click into my total impressions to see what's happening. Now right here, I see that impressions have increased week over week over the last. And this is amazing because I get to see that it's increased based on historicals and based on the expected range. I can break it down by region. And even better, this is the coolest part. I can even ask questions. So again, can you imagine? Now you can actually ask questions about your data. So I can say, all right, what are some of the questions I want to ask? I want to learn more about what's happening with my impressions. Okay? So it's going to even give me questions I should ask to learn more about my business. Now, all of this would take how many days? One, two days from your team to be able to dig into what's happening? Now I can get it. Literally, you guys are watching with just a matter of clicks. And even better, I can share this with my marketing team. So now the marketing team can take action on this, right? They're seeing that impressions are increasing in apac now they can adjust their budgets accordingly. So just like that, I'm taking action on my data. This is why I love Tableau Pull so much. Again, it's giving me the opportunity to truly democratize AI insights so that I can action on my business. Now, I just talked about step four, right? All about analytics with AI. Now I'm going to wrap up this keynote by talking about the fifth step. And this is all around how we could better collaborate with AI. Now, as a marketing leader, I'm sure a lot of people in the room can also feel the pain of how many interactions we have to juggle. You have interactions with your internal teams. You have interactions with vendors like agencies, customers. A lot of the tools that we use, like email and text messaging, really isn't that efficient, Right? So wouldn't it be nice if there was a Better way that we can really collaborate with all of these people that are so important to our business. Well, this is where the power of Slack comes in. So Slack is truly the central hub. It brings together conversations, it brings in collaboration, it brings in automation, and now it brings in the power of AI all in one single place, truly changing how we work. And the Slack team, just like Tableau, has been working really, really hard on lots of new incredible AI innovations. So I'm going to quickly tell you about them and then, better I'm going to show you. So the first one is Copilot. Again, all the same exact copilot, right? That incredible intuitive AI assistant is now available in Slack. So if you're in Slack, you can talk to your CRM data the exact same way that you talk to your colleagues, allowing you to avoid the swivel chairing and allowing you to actually move forward, forward with work, right in your conversations. Okay, record channels now. This is incredible. It's taking all of the record data and it's putting it right into Slack so now you can collaborate cross functionally, right on the most important opportunities and deals. As a marketing leader, I don't really get a lot of access to my CRM data and so now I can sit side by side with sales to help close the most important deals to add more value right in the moment. It matters. Okay, sales elevate. Any sales leaders or anyone that does sales in the room? Okay. Few people tell your sales friends this for sales elevate. It's amazing. It's bringing in all the sales cloud data right into Slack. So again, you get real time deal alerts, you get to automate workflows and actions and you can edit all of your CRM on the go from any device. And then finally, Slack AI. This is incredible. I can't wait to show you. This is allowing you to search and summarize everything, all of the workforce conversations. It's truly tapping into the collective knowledge and history right in Slack. Enough of the telling. Now, I am so excited to show you the final demo. This is the demo of all demos. It's bringing together everything you've seen all through the power of Slack. Okay, so this is how Aston Martin uses Slack. Now, question, I can get a little bit frustrated. I know. I talk to customers every single day. It's kind of amazing how much time it takes just to get going in the morning, right? You wake up, kiss your kids, have a cup of coffee and then you have to dig into your email, dig into your text messages, call sales Leaders like Stephanie Glenn right here, Right. Just to catch up on what happened while you were sleeping. That is insane. But now with Slack AI, that is no longer because we have recap. So check this out. Recap is literally sending you a daily digest of all the mentions, all the conversations, all the activities that you missed while you were away. So right here you see marketing strategy. I had just sent that tableau Pulse Insight when I was up at the demo desk. And you see there's a lot of positive reaction from the team on this insight and they want more information. Now while recap is amazing at summarizing what's happening, I want to dig into the data now before I have to leave the Slack channel, I have to go into a data set, find the answer, go back into Slack just to send it. Now I don't need to do that. Why? Copilot, Copilot is right here with me, right in this channel. It has all the context to this recap. I can literally type in what is the most popular customization in this region. And just like that, I get the information. This is incredible. So I can share this right back out to Team Japan, give them the question they are asking and they can take action on it again right in the place I work. Okay, let's go back to the recap. Let's scroll down. Okay, stop right here. Do you guys see right in the Aston Martin deal we just closed, we got a deal win alert. Now this was an incredible five day nail biter. Has anyone ever been in deals like that? Maybe.

Yes.

Okay, so I'm going to walk you through exactly what happened. Let's click into the Opportunity channel. Okay, amazing. Now, any Slack users? Yes, that is amazing. I love Slack too. Okay, but does this for the Slack users? Does this look a little bit different, this channel? It does, right? This is the record. This is the Record channel, right? So this is where you get all of that rich GRM data right here, right in conversation. So with the Record channel, I can edit, right? I can automate workflows and even cooler, well, not cooler than that, but I can also bring in amazing Slack features like Slack lists. So Slack List is going to let me collaborate again, cross functionally to prioritize what is most important to the business. How do I need to close this deal? Okay, let's go back to the messages because I'm telling you, it's a really good story. Okay, so a customer, a dealer reached out and said, I have this really bespoke customer request. I need help. The customer, Amy, she wanted to have the interior stitching of her car match the Chicago blue. That is like very bespoke. Right. And while that's cool, I thought Chicago would be psyched about this. That's pretty cool right now. Bespoke color, very rare. And so if you scroll up, you can see that that incredible model that Michael built, that propensity to buy model. Well, real data came in and said, actually we're seeing that the propensity to buy went down. So Amy is probably not going to buy the car for two reasons. It's actually giving you the reasons. The first one is because it's such a rare color, the supplier needs a little bit more time to secure it. So now we're not going to get the material until 2025. Number two is it's a rare color, so not a lot of suppliers have it. That's okay because Copilot again is right here in Slack taking action. It is 10 steps ahead. So it is automatically going to send an email to every single available supplier to see if there's an alternative. Right. This is pretty incredible. Now I want to, I want you to take a quick look at all of those. You see that conversation history? Look at all of those emails. Can you imagine if that core sales rep had to do that repetitive task of the emailing and the responding and the emailing? Now that rep was able to focus on the high value sales rather than the repetitive tasks that Copilot took care of. Okay, so going back to the story, 15 or not 15 days, like 15 days, three days later, the deal was closed. Amy bought the car and she was psyched. Now it sounds like a lot of magic. So I just really want to reinforce what just happened here. Number one, Michael built that model. That model was not just for his specific use case. It's pervasive. It's across the entire enterprise. That model was used. Real time data came in and the data dropped the score. Copilot proactively reached out on Slack to let us know that action was going to be taken. Number two, when the score dropped, it triggered an automated workflow. So that workflow sent Copilot to send an email to all the available suppliers. That's incredible. And when Copilot secured a supplier, the opportunity record was automatically updated. Can we get a round of applause? Does that happen in your workforce right now? No, that's Copilot. That's amazing. So that is the power of AI in Slack and that is the power of AI in action. All right, Ariel, back over to you.

Kelly. Thank you. That was fantastic. Well done. Okay, so we learned how to build a customer360, how to unify data with data cloud. Michael showed us how to use Einstein Copilot to have AI in the flow of work directly in your favorite Salesforce applications. And we talked about having AI make analytics better to really let us talk to our analytics as if it was an analyst and our amazing Slack AI features. Well, this is the start of your journey at Connections 2024 and we have an amazing conference for you with a ton of great content. 270 sessions, 50 demos. I especially like to call out the Sessions with my MarTech team if you want to learn more about how Salesforce is deploying marketing technology. Thank you. There are a lot of them are here and great customer stories as well. But one last request I have as a marketer, I would like some data from you. So we love feedback from our customers. If you could scan that QR code and give us some feedback on our keynotes, we can help make it better. We would love to get that and you have a chance to win a free pass to Dreamforce. So thank you everyone. Appreciate your time and I hope you have a great time at the conference. Thank you.