Skill Exchange Keynote - The Future of Experience Making with AI, Adobe Experience Platform

Transcript

Good morning, good afternoon, good evening, or maybe good night, and thank you for joining the Skill Exchange. I’m here to kick it off for you. My name is Clans van Tooker, Senior Director of Product Marketing for Adobe Experience Platform. I’ve been with Adobe for over 17 years and working for the past seven years on the Experience Platform initiative. It’s been really exciting to see what you have been doing with the Adobe technology, but it’s also a little bit interesting and maybe even scary to see what is ahead of us. In this keynote, I want to walk you through my observations, what I see in markets, and what I think you can learn during Skill Exchange, or areas of investment that make you better as an experience maker or better with our tools and technologies overall. So let’s have a quick look. What is happening in our world? Well, it’s probably also something that you’re seeing. So let’s start first of all, we’re building experiences for consumers and businesses. And data, understanding consumers is key of what is happening. But one thing we know for sure is that the amount of data that we know about our consumers is ever increasing. And how do I make basically the selection between what is relevant and what is not? That’s a hard job. And a fun joke that I always hear is for a data analyst to understand the consumers better and searching a needle in the haystack, they’re asking for more hay. But more hay doesn’t always mean better signals. So this is one of the challenges that we need to work with on how do we distill better signal for noise. The second challenge is consumer attention span really shortens. Just look at our own behavior. We get a push notification on a phone and our attention is distracted. Maybe during this keynote presentation already, you’ve got multiple beeps and bings on the various devices, and your attention span is shortening. So how do we deliver really that right message, that compelling way to engage and keep that attention span just a little bit longer? And the time of broadcast, the time of everyone receiving the same message is way behind us. One of the unique opportunities of this one-to-one engagement is the ability to personalize it. When you meet with family or friends, you exactly know what’s top of mind for them. You can speak to them related to what they’ve been experiencing or what they’re going to experience. And that’s why you see this board of all the pictures, because different people have different interests and how do I basically deliver the right content with the increased needs for content personalization and the content variance? That’s not an easy job.

And then lastly, we are not getting the same amount of time to actually build the experiences. And we’ve been asked by our managers, by our bosses, by our companies to do exactly the same thing or even better in just less time. And this doesn’t look like a great equation, but actually this equation has opportunities as well. Because if I look at what is happening, especially in the current age, Agenetec AI is taking the world of digital experiences and taking the world of digital by storm. And what are the opportunities to talk about the adoption of Agenetec AI in how we are changing our workflows and our approaches to automation? And further on in this keynote, I’m going to show you a couple of examples on what you can do today and what is coming in the near future.

But what is actually top of mind for practitioners like you? We spoke about the challenges, but as a practitioner, number one, of course, we want to deliver on the strategic objectives of our employers, of the businesses that we work for. That’s great, but we also have to think about ourselves. And when we think about ourselves, it’s the continued interest to learn new skills and new technology.

And those new skills and technology is something that you’re picking up here. As Adobe, we want to make sure you have the right skills and the right technology. We want to make sure you have the right knowledge. And that is why experience makers, and in this case, the skill exchange is such a unique forum for you to be inspired and learn.

Second is sharing experiences, because sitting in a classroom, whether it’s virtually or live or going to a conference is great, but actually step up, share those experiences, which increases your street credibility as an experience maker. And this is where you will see some great presenters from your community during this skill exchange.

And lastly, by doing that, you’re growing your own value. I’ve been in this business now for over 20 years, and I’m continuously trying to learn and continuously trying to grow my own value, not in this community, not in the company, but in everything that I want to work with. And I think that’s what you’re trying to do as well. So what is the biggest challenge that I see today? There are so many opportunities, there are so many things to do, but how do we go from a blank page to an in-market campaign or an in-market experience as quickly as possible? The blank page syndrome is real. We can throw more technology at the problem, but where do I start? How do I get inspired? And maybe that’s why you’re here today as well. Maybe that’s why, let’s say you attend other activities and events from Adobe or other vendors or for example, Adobe Summit, to be inspired. And one of the examples that I’m trying to tell people about is if you’re using Adobe or one of the applications, there’s a use case playbook as repeatable recipes right within the software. It helps you to have a conversation with your business stakeholders, have a conversation about what do you want to do. And we always might think that what we want to do is unique and one of a kind, but actually they follow similar patterns. So start diving and talking into use cases and the conversations with your business instead of just starting with the blank page.

And a lot of research, but also working with our ultimate success teams is that having great technology and having great use cases is great, but actually influencing your organization, having the right organizational readiness and the blessing and the support from your leadership is key. So don’t only think about the technology or only think about what is delivered, but work your organizational skills to basically set you up for success. And lastly, acceleration for AI. In the engagements with my teams, I’m actively encouraging them to use AI, to use AI tools like AI assistants, like Microsoft Co-Pilot, like JetDPT on a daily basis. It’s not a muscle memory that everyone is learning right from Scott. Actually, very few of us on this call right now start to do this in school. Yes, we’ve learned a lot in school. We’ve learned math, we’ve learned language, we’ve learned science, but actually the ability to prompt into a system and get an answer back is something we have to really put energy into to learn. And I think that is my simple call out to you guys. Spend time in prompting because prompting is the art of expressing what you want. For example, recently in a discussion with my wife, she said, hey, Katie, can you go to a supermarket and bring four cartons of milk? And by the way, when they have eggs, please bring six. So I came home with six cartons of milk. And I know it’s a horrible joke, but it speaks about the articulation and the importance of prompting. So to take it from the concept more into action, I want to show you what experience platform AI assistant can do today in the Adobe technology. So let’s start with the first one. Data insights agent. Customers using customer journey analytics can use data insights agent to explore and work with data without clicking any mouse button. You see that the data analyst in this case says, compare the orders for me by product category for September and October 2024. It’s just how you would ask your colleague. And the system is interpreting and comes up with this visualization right out of the box.

Then the next one. In experience platform, you can build a lot of audiences, but which audiences actually have a large number of profiles, which audiences have a small number of profiles, or actually which audiences have gone down to zero profiles over time. And here you see an example created by my colleague, Danny Miller, and he wrote list any audience with zero profiles, the creation date, the first date the audience reached zero over a month ago. And you see that this natural language prompt is being translated into a query into the system, which is actually more difficult to build by itself if you would be using the out of the box user interface.

And then actually, where are we going? And one of the things that you see is you’re using products like customer unit analytics, AEM, Adobe campaign or journey optimizer, and you’re switching between those products. But in the world of AI, you will see these experiences coming together in a much more fluent way. So I will share you a demo that is coming straight from our engineering environment. But before I do that, let’s look at the last prompt. This is a prompt that you actually can run today.

We have support for streaming audiences. List all the streaming audiences, their population, the count of used and other audiences, count in destinations, and the count of sum of journeys. It literally looks at all the audiences used in the system. And yes, that single language was constructed in this very complex technical prompt that you now see on the screen. With that, let’s lift up the curtain a little bit. And I’m going to give you a preview of Experience Platform Agent Orchestrator and things we’re looking to release in September or October of this year. And please don’t tell anyone yet, except for your colleagues. This is really exciting. So let’s have a look. I’m a marketing specialist at a large outdoor company. I’m planning a new campaign for weekend hiking. And I want to highlight some of my products that customers may not have bought yet. Ones that could be used for both backpackers and day hiker. Let’s see how AI Assistant can help me build a personalized promotion. But before I get started, let’s analyze some data. I ask AI Assistant to show me the revenue by product subcategory for camping in June. I want the top 20. AI Assistant quickly pulls in the right information from Customer Journey Analytics to show me purchase information for my top products from June. It’s a bit of a long list, so I’ll ask for AI Assistant to help me narrow this down to the products that can be used for both backpackers and day hikers. That narrows things down for me. I’ll focus on these products when we get into the details of my campaign. So let’s get started. Please help me plan an email marketing campaign for the fall. My focus will be to target weekend hiking. I want to experiment with two variants.

Once I submit this prompt, AI Assistant comes back with a full plan on executing the campaign. Now let’s take a minute to unpack what we’re seeing here. The first thing I see is AI Assistant Reasons to make sure that it’s understanding my question and uses that to come up with a complete plan. The plan has a goal as well as four different phases, including creating audiences, setting up the journey, generating new content, and setting up some monitoring and scheduling. Each phase also has certain subtasks and statuses. I realize I missed out on a few details on the audience creation. I’ll add some audience specifications, including some of the information from earlier in the conversation. As you can see, it’s gone ahead and added a new constraint to my plan regarding the audience size. One last thing before we proceed. I want to remove the monitoring phase for my plan. I think I’ll just do that later. Once again, AI Assistant has quickly adjusted the plan. The updates look great. So let’s proceed with the first phase to build a targeted audience.

I see that AI Assistant has gone ahead and performed a few steps of this phase, including discovering XTM fields, translating that to an audience definition, and estimating the audience size. The audience size is in the right range for the constraint that I’ve given it for 10,000. So I’ll go ahead with this audience. You can see here on the right that this has gone from in progress to complete. It created the audience, and I can always access it on the audience portal to make any additional changes if needed.

And now that phase one is complete, I’m guided to the next phase of our plan to configure and create the email journey. Let’s proceed. AI Assistant responds with the specifications of the journey that it recommends I create. As you can see, this is a pretty complex journey, including certain branches for the experimentation as well as entry and exit criteria. It also shows me the journey specifications and natural language for my review, making these technical details easy for me to understand. It’s looking good. Let’s go ahead and create the journey. AI Assistant comes back with a fully created journey and shows me the name and the unique idea of the journey that’s created.

I can make edits from here or go directly into Adobe Journey Optimizer and make edits there. The last phase of this plan is to set up content. I ask AI Assistant, what content variations should I use for my weekend hiking campaign? AI Assistant analyzes past performance of my content and picks the top themes that it thinks will perform well. These look pretty good to me, so I’m going to ask it to generate two variants.

AI Assistant uses the previous recommendations and the Adobe Firefly models to come up with two different content variants to use in my email campaign. Both of these variants feel very on theme for me. I think they could really perform well for my hiking campaign.

As you can see now, all three phases of my plan are now complete and I think my work here is done. Thank you, AI Assistant. Whoa, that was a lot. Because what you’re seeing here are the things that we spoke about. First of all, working with a large amount of data to understand through a natural language what is the audience that we want to use. Number two, using AI to basically help express what we want to translate into an actual campaign. Number three, using those personalization elements and creating the content variants as well as have the content created for me. As you can see that, the time spent that it took is significantly shorter. I know this is not ready to completely roll out. We need to test, we need to refine, but imagine going from a blank page to this with the traditional way of working. This is going to be totally different. And yes, I’m really excited about this and this is coming to your Adobe solution real soon.

So let’s talk about next steps because we’re here at the Skill Exchange and what are things that I recommend to look at first. First of all, embrace agentic AI. Start expanding your prompting skills. There are a lot of tutorials, examples, and tools out in the internet to improve prompting and improve the position of us using natural language with a machine to improve the outcome.

Second, learn AI assistant in the Adobe technology. It is going to be the interface for you to interact with Experience Cloud in a much agentic way and the things that you just just shown in the demo.

And lastly, share experiences. You’re here at the Skill Exchange today and maybe you’re listening today. If you want to be a future presenter or maybe at Adobe Summit or any other event, sharing the experiences not only gives you energy, but also increases your street credibility.

And with that, I really want you to enjoy the Skill Exchange. It is a unique opportunity to learn from your fellow members in the community. Have a great event. Thank you so much for sharing and for being with us today.

Thank you, Nimesha. I’m really excited to be here with this global audience. I just got some cues from the team producing this on the back end and we’ve got people as I understand from all over the world. So from wherever you’re dialing, good morning, good evening, good afternoon, or even good night.

Thank you, Katie. So this is one of the first keynote session Q&A. We are facing some difficulties with our Q&A function. So guys, if you see a box and you can submit the question, please do so. We’ll make sure to get them answered. For now, we’ll discuss some general questions. Let us keep the conversation moving.

So this is a practitioner based Skill Exchange. So I’m asking one of the practitioner questions. As a practitioner, what new skills should they focus to stay relevant as agent orchestration become more common? How do you foresee the role of a practitioner? That’s a very good question, Nimesha. And when your team contacted me to say, are you interested to present at Skill Exchange and what would your message be? This was actually one of the things that is top of mind. I like to be forward looking. I like to give you guys a heads up on where I think things are going. And that’s why you saw that very heavy focus on jobs to be done, but also the influence of AI. And if there’s one thing that I can recommend everyone for diving into, it is actually mastering, let’s say, AI capabilities. And it’s amazing. Yeah I’m subscribed to a number of newsletters. I follow podcasts and other things. And actually, I feel overwhelmed. I literally do not have enough time to listen to it, to catch up on it, or even to try it out. So it’s a whole new thing in working. When I started, let’s say, in this domain, I learned computer programming. I learned in new programming languages. I learned web technology. But now, let’s say, learning all this AI stuff, learning to write better prompts, learning to understand what is possible, I really suggest everyone to be diving into that and understand what’s possible. Because to me, it’s not a matter of if, but it’s a matter of when this will be coming to any of the technology we’re using today. And especially, of course, as I showed in the session, Adobe technology as well.

Thank you, Katie. It was insightful. Since you talked about organizational readiness, so are we looking to provide some blueprint for organizational readiness or be agent ready kind of playbook? Yeah. And let’s step back, Nimesh, for a moment from that question. And what I mean is a lot of our customers license the technology and then look at it as how do I start.

And this forum, Skills Exchange, of the larger forum, Experience League, is already a very good, let’s say, forum to be participating in. What are you doing? What are you learning? Where do you start? Yeah, we from Adobe, we provide stuff, but actually coming and seeing it coming from the community about what works, how does it work, what’s the idea that I had, what did I brought it to life. But we also want to bootstrap. So for example, if you take AI Assistant, of which I showed a demo in the presentation that I just did, there will be example questions. But we’re pushing the boundary on our agent orchestrator and agents. We’re pushing the boundary of what is possible. And we want to inspire you. So whenever you see something showing up on Experience League or in product or at an event, try those things out. Build the muscle memory because the white page syndrome, let’s say, is the biggest challenge that we have, like great stuff, but where do I start? So true, so true. I completely understand that. So moving toward next question. It’s towards, I would like to let the audience know behind the scene. So can you help us understand what is happening on the back end in agent orchestrator and break down the complex questions like you showed in the demo earlier? Yeah, let’s do that for a moment. And I will take an analogy. Probably something we do at least once a week is to cook a meal. So we have the pots and pans. We go to the supermarket to buy the ingredients. And somehow we’ve learned how to do that. Maybe we looked at the cookbook or our parents or grandparents showed us how to do that. And that is how we learn to combine both the ingredients with the tools that we’re using. And agent orchestrator is basically doing exactly the same thing. It has the tools. It’s your experience cloud solutions. It has the ingredients, which are the audiences, the journeys and everything that you’re happening. But it’s also being fed the information that sits on experience leak that is specifically to our domain. Because how would a system know what is a journey? What is an audience? What are the top buying customers? So when you enter a prompt into AI assistant, it’s being picked up by agent orchestrator. Agent orchestrator first tries to understand what are the components that you’re asking for and tries to put that into the world. And that is using a large language model on the back end. That is generically trained. No specific customer data is used in that context at all. Just to understand the question. Then from that question, it will break it down into a strategy.

In the example that you just saw, the first step was like, I need to understand a top customer segment for a particular product. Those are typically questions that are combined and coming, for example, from customer journey analytics. So it will know how to frame a sub problem into the data insights agent.

Yes, I know it’s a data insights agent on the back end, but that’s totally hidden and you shouldn’t have to worry about that. Once you have that, the next step is like, okay, I have this audience. Now I want to turn, I have this group of people. Now I want to turn it into an audience. And that is where you see the audience agent playing out. So just like we do as humans, when we are given a complex problem, it’s broken down into little pieces. Then there are executed step by step by step. But the great thing about what we show is that you as a user are in the driver’s seat to do, let’s say the steps that you want to do.

Thank you. That was really insightful. And we have Dushar who is asking us, can we make journey by asking AI means internal AI can be able to design journey, right? Well, yes, exactly. That’s where we’re going and what we showcased in that quick preview. And by the way, the preview is really, really close. It’s just I think, one or two more months that you see this showing up. In the example I showed for a journey to be created to target enthusiasts in a particular product domain. So this is exactly where we can help you build a journey. That doesn’t mean Dushar that that journey is actually the final journey that you’re going to create. But it’s helping you to get started and do the basic work that you would normally take, let’s say minutes to hours to create that. Thank you. And we got very interesting question that is towards the use case based. So will AI can design the content for a personalized email, say cart abandonment use case, email will with users cart products. Also will AI assistant can help in designing personalized landing page content pages? Oh, yes, absolutely. If you think about journey optimizer, it is serving customers both inbound people that land on a web page and probably need a personalized experience in their own language versus outbound. What are we doing from a push perspective, push connectivity from a in app experience or even from an email perspective. And as we build this out, you as a practitioner should be able to ask for what you need. And for our system to put together that experience independent whether it is an email or a corresponding landing page, or even better, the two that are tightly connected and will give you exactly the same experience independent of whether the email was received, or a customer would go to the landing page. So that’s absolutely where we’re going.

Thank you, Katie. And we got a very interesting question. Most of the guys think about it. Are AI tools included in AGO purchase? So very, very good question. And with these AI tools, we will see a slightly different go to market than you’re used to from Adobe. Any customer who has the right, let’s say, what I call legal framework, there is some potential additional license terms that need to be agreed on, will get access to these capabilities. It’s basically a one time, let’s say, legal document to allow for these capabilities to show up, which we call the agent orchestrator capabilities. And at that point in time, there’s going to be a consumption based approach. We assume that by doing this, we deliver value with every prompt with every question that you will be asking. And in the backends, we’re going to a credit based approach. And no, we don’t want you to make a decision every time that you fire off a prompt, do I want to do this? These are such large volumes against such small amounts that this is handled within your enterprise at large scale in the same way as other companies with prompt based AI systems are doing that today. Thank you, Katie. It was again, insightful. And we have a very last question now. So it’s towards privacy, people are often confused around it and want to know more. So here is the question.

In the context of Adobe AI systems looks to me, prompt engineering is very key. What kind of tools, BKMs, you, Adobe recommend to get started in agentic AI? There are concerns in GDPR law countries related to AI agents crawling and consuming the data. How do I explain to the business that solution is complaint in terms of protecting the AI data? That’s a very good question, because we want to know as brands, what is happening to our customers? Is it used to train my data? What is, is my data going somewhere else? First of all, what you do in the customer context stays in your context. We are not using the data to train our systems or to train basically what is happening in the broader experience cloud domain. What we do, however, is we observe the prompts that you’re typing to understand if the right output is given, but that’s only in the context of your system. And if you want to know more, please go to Experience League. Please read up on what is happening in AI Assistant. And we should be able to help you get across these little, let’s say, I’m not really comfortable yet help me get there. But I think that’s natural with any new technology and new technology adoption. So I know time flies. Last question, but Nimasha, thank you. Great questions and insightful answers. Thank you so much, Klaasjean, for having me here. Thank you, Nimasha, and a great continuation of this event, everyone.