Agents at work: Creating and activating audiences and journeys

Transcript

He finds the hidden faces, spots rises sudden falls He hunts down all duplicates, cleans chaos from your walls He saws his reach before launch, turns questions into sets Then creates the perfect audience, behold audience agent Oh Agent Orchestrator, mighty hand is shining screen You can dock a thousand agents, like an on-scene machine He maps the branching pathways, times pauses, starts and flows He spots the quiet drop-offs, where every journey slows He tests and choose the pace and finds vaults you can’t quite see Then builds the perfect journey, behold journey agent Oh Agent Orchestrator, mighty hand is shining screen Why are you saying I can’t hold it all? All the threats and all the timings, every morning, every change In the quiet sun just thinking how to shield the real race Let the agents take the burden, let the bloods unfold unseen While I’m here, the mind keeps working in the heart of the machine The agents are gonna save my sheen In the quiet sun just thinking how to shield the real race He finds the hidden faces, spots rises sudden falls He hunts down all duplicates, cleans chaos from your walls He saws his reach before launch, turns questions into sets Then creates the perfect audience, behold audience agent Oh Agent Orchestrator, mighty hand is shining screen You can dock a thousand agents, like an unseen machine He maps the branching pathways, times pauses, starts and flows He spots the quiet drop-offs, where every journey slows He tests and choose the pace and finds vaults you can’t quite see Then builds the perfect journey, behold journey agent Oh Agent Orchestrator, mighty hand is shining screen You can dock a thousand agents Hey, welcome to our latest episode of Experience League Live Agents at Work, creating and activating audiences and journeys I’m your host, Daniel Wright, Senior Technical Marketing Engineer Confess I’m feeling a little extra senior this week Got a big birthday coming up on Groundhog’s Day Wish me luck getting through another trip around the sun So we’ve been doing a series on Adobe Experience Platform Agent Orchestrator Over the last few months We’ve had episodes on Orchestrator, Product Support Agent Today we’re excited to continue this series with this episode covering Audience Agent and Journey Agent This episode is going to focus on B2C use cases and an upcoming episode will show off some B2B scenarios Before we get going, I have a question for all of you in the audience We like to have dialogue during Experience League Live So enter comments, questions in the chat I’ll get us started off with this question If you would like to post how you are using Gen AI or Agenic AI We’d love to hear your responses and get the conversation going Me personally, I’ve been having a lot of fun outside of work I’ve been having fun generating songs with AI to accompany each one of these Experience League Live episodes Whenever we talk about AI agents, there’s always someone who can’t resist making a 007 joke So that was my homage to some of the great Bond theme songs Alright, so let’s welcome our guests If you’ve watched Experience League Live before, we usually show we have fun facts for each guest This month is going to be some surprise fun facts So let’s welcome our first guest, Senior Manager, Product Management, Real-Time CDP Welcome Shelby Monsouri Good morning Good morning Shelby’s fun fact is that she held the rope on the Snoopy balloon at the Macy’s Thanksgiving Day Parade So how was that experience Shelby? Holding the rope? It was out of this world, I can’t even describe to you what that experience felt like But no, it was one for the memory books Yeah, and it’s almost like coming into work on a holiday too Because it’s right near the New York City Adobe office Yeah, that’s right, Daniel and I are both based in New York We’re a stone’s throw away from where the Macy’s Day Parade balloons set up So it’s kind of fun, you can go check out the balloons the night before Thanksgiving So I’ve done that before as well as holding the rope I’m kind of a Macy’s Day Parade veteran That’s great, that’s a big job and we know you can handle a lot of responsibilities Shelby So well done Thank you Our next guest is product manager for Journey Optimizer Welcome Kira Fawcett Thank you, Warren And Kira’s fun fact is that she can play any song on the piano by ear But she can’t read music So that’s a pretty incredible skill, Kira When did you first realize you had this talent? I was listening to Spotify one day and figured out that I was just going to go Well sorry, I decided I was going to go try the song And then I played that one and then the next one And then I practiced a lot and then I got better and better and better And now I can do pretty much any song except for some will take a little longer Like classical music will take me a long time That’s amazing That’s a song from Charlie Brown Just for you We can accompany Shelby next year on the parade route That’s amazing though, this is a talent that you discovered when you were an adult Yes, so my dad actually has the same talent And so I was kind of jealous of him I think and so I tried it out And I had piano lessons as a kid, I never really practiced very much Learning how to read the music But something stuck with me so yeah It was an adult acquired skill I guess That’s amazing, I’m going to have to try this when I get home See if I can see if there’s any latent talent there I do have, well I actually do have one talent related to that Any song with clapping in it, I can do the same clapping without reading music So I’ve got this one We Will Rock You Queen And I’ve also got the Friends song There you go Ah We’ve got talents Alright let’s welcome our third guest, Senior Product Marketing Manager Alyssa Espiritu Welcome Alyssa, and Alyssa’s fun fact is that she’s currently working on getting her Pilates Teacher Training Certificate How’s that going, how much more time do you have left? Lots of hours, plenty of hours left But I will say it’s a lot of fun Clearly I’m not as talented as Kira, not as strong as Shelby To have to hold down Macy’s Parade balloon That must have been heavy Shelby What did you say? I said that must have been pretty heavy Well so as many hours as Alyssa has spent training for Pilates instruction I also had to do some strength training for my rope holding She has very strong arms She’s very helpful if anyone is moving within the New York City area She can carry some boxes Alyssa, my wife teaches Pilates and you know what I predict your next certificate will be in? Oh no, what is it? Therapy, therapy There is so much talking that goes on when my wife teaches Pilates And that talking has nothing to do with anyone’s core It’s like every topic is discussed as she does these private sessions with her clients That is true Alright so let’s get going with the episode, shall we? So first up is Alyssa who is going to share some background on these agents Yeah, absolutely, thanks Daniel So thank you everyone for joining It’s great that we have such a big audience today I’m sure you’ve heard about all the amazing stuff that Adobe has released As it relates to our AEP agents and our agent AI, agentic AI strategy But if you haven’t heard about AEP agent orchestrator and our agents within our platform I’m just going to give you a little bit of context behind that So that when you see the demo you’ll have a little bit more understanding of how and why we built these agents So our agentic AI capabilities at Adobe and within Adobe Experience platform Are really built for our marketing users So that you guys can execute on customer experience orchestration So we aim to remove bottlenecks and hopefully even increase your capacity And your organization’s capacity to enable your teams to focus on more unique And important strengths for your business such as creativity, marketing strategy, business insights And ultimately delighting your customers So agent orchestrator provides this foundation so that you can unlock the full potential of our AEP agents And our agentic capabilities in your day to day roles and your day to day workflows And you’ll see what that looks like within the UI And how you can apply that to what you’re already doing today So we have built agent orchestrator natively on top of Adobe Experience platform As you can see in the bottom of that screen right here And it allows us to deliver agentic capabilities that take advantage of the shared data infrastructure of our platform While still delivering trust, transparency and the appropriate security and guardrails that you need for your business So we have mapped all our agents to the core functions of the different customer lifecycle that drive customer experiences So each agent that you see here is purpose built to address specific operational problems and operational jobs So from planning and analysis to audience management which you’ll see later today that Shelby will show you Content production and all the way to journey orchestration which is what Kier will show you So these are not just general purpose assistance per se But they’re really specialized to align to how your marketing teams work across the entire organization And so as Daniel said earlier today, we’re focusing on B2C use cases today We also have B2B use cases that’s for a different show But our agents are really designed to move forward and live within your existing workflows And work with you and how you do your roles to help you supplement your day to day role And we don’t just surface insights, they’re really working alongside humans in that workflow with an AEP So what you’ll see today and you’ll see a couple of these skills, agent skills in the demo You’ll see audience agent which is available in real time CDP and Adobe Journey Optimizer It’s really designed to tackle some of the most persistent challenges that marketers face when you’re building audiences So today many teams might be stuck manually configuring all these complex rules It takes a long time to diagnose issues sometimes but creating audiences is also time consuming And we’ve also seen sometimes onboarding especially if you’re a new user could be a little bit overwhelming But these hurdles lead to inconsistent audience quality and slow time to market when these challenges keep on persisting So we created audience agent to help expedite that audience management and exploration workflows We also created audience agent to help accelerate how you can build and create audiences And we also want to help users be more informed when they create audiences so that you can build more targeted and personalized audiences as well So in the demo you’ll see Shelby work with audience agent to discover audience fields and get a better understanding of existing audiences that she has in her environment You’ll also see her create audiences in a more streamlined way working alongside the agent Coming soon is that audience creation skill that I just mentioned And we’re also working on the ability to generate propensity models as well later this year So a lot to look forward to there Great And then lastly, you know, before Kira and Shelby do their demo, I also want to talk about journey agent This is also an important piece that can kind of go hand in hand with audiences and journeys and real time CDP and journey optimizer, right? So for journey agent workflows, you know, our agent really works as that subject matter expert for marketers within journey optimizer So we created journey agent to which is designed to make, you know, building and optimizing journeys easier and hopefully a lot more effective, right? So journey agent helps identify audience overlaps, scheduling conflicts, which really enables marketers to kind of arbitrate between journeys and prevent message oversaturation Which is what we’re seeing a lot, especially among, you know, marketing So you’ll see Kira talk a little bit more about how, you know, you can analyze the certain journeys that you have going on today And then we also have drop off analysis where agents can pinpoint where customer profiles may fall off at certain points of the journey And then we have journey agent that provides, you know, better insights and maybe actionable recommendations to inform marketers on how you can boost engagement and conversions by understanding where the fallout is Journey insights also allows marketers to ask questions like show me all the live journeys that I have today And also surface operational insights, making it easier to manage your journeys And you’ll see this in the demo as well. But coming soon, we will also be able to generate and draft journeys, multi-step journeys from text based prompts So this allows for, you know, more streamlined ease of use and intuitive way to create your journey So with that, I’ll kick it off to our amazing audience and journey specialists to show you a demo to see what it looks like in the UI Great. Thanks, Alyssa. And over to Shelby Actually, someone just asked if this is a recorded session. Yes, this is being recorded and it will be available on experience league after the show Awesome. So let me get into the demo. I’m going to start with a general overview of what you can expect when you start with AI assistant and the agents powered by agent orchestrator And then we’ll get into a real world use case with audience agent and journey agent So as you can see, I am on our experience cloud home. I was able to navigate to AI assistant directly from here without specifying a specific application with an experience cloud But I am able to access AI assistant from anywhere. So this is my home screen for AI assistant. But once we’re in the tool, I’ll show you how I can access it from a specific application So I’m presented with the chat bar to ask my first prompt. But we give you a couple of tips and tricks right off the bat on the home screen So there’s a series of suggested prompts that are related to different agents that are currently live. You can see for audiences, there’s a suggested prompt to show fields related to email Or if you have product knowledge questions about audience management, you can start directly from here. There’s also a prompt Hey Toby, before we go deeper into the audience agent, I just want to ask, so I see you’re logged into experience cloud and you have this AI assistant in the left rail Is this something every experience cloud customer would see when they log into experience cloud or is there some permissions or special privileges that are needed to access this? Yeah, great question. So AI assistant as a capability is available to all users logging into experience cloud But to get access to this specific experience where it’s showcasing agent agenetic skills, you would need to talk to your account representative to route that Okay, so if you don’t see what Shelby sees when you log in, maybe first speak to an administrator of Adobe products at your company and ask them if it’s a permission If you just don’t have permissions or if it’s not just a permissions thing, that administrator would need to talk to their Adobe representatives about getting access to the capability Right, and Alyssa can share more details in the chat about what the next steps would be there, but just to clarify what you called out Daniel, it’s not a permissions toggle where we turn it on and off, you would need to go through your account representative for now to get access to it And then you would also need like RTCDP or journey optimizer Yeah, so AI assistant is aware of what your application and user permissions are So any role based access controls that are set for you. So in my case, I have access to a bunch of applications If I wanted to invoke, let’s say I was going into CDP and I wanted to invoke AI assistant, I have it available on the homepage I can also invoke it here in the top nav. So as I’m navigating CDP, I can also talk to AI assistant and leverage some of the agentic skills So it’s available within each application through the top nav. And then if you’re not within an application, but you have questions Or you have a skill that you want to kick off without having to go directly into that application, you can do that from here Great. And Yash in the chat is asking, I guess, is in case they reach out to their Adobe representative What’s the exact name of the capability that they should ask for? Is it agent orchestrator? Yeah, I think agent is the buzzword. So they’ll know they’ll know where to send you if you ask for access to agents But agent orchestrator is a great first pass. And then if you have a particular interest in audience agent or journey agent Or any of the other agentic capabilities that we talk about in these shows, then they can route you to the right person Great. Thanks, Shelby. Let’s start to slow you down No, no, it’s great. Good questions. And I’m also it’s great to have the questions, Daniel, because I’m so familiar with what I’m seeing that it’s good to have that call out It feels I don’t want to assume anything. One call out for when you’re accessing AI assistant and what kind of, in my opinion, is the beauty of this product Is that the application silos are not there. I’m in Experience Cloud Home, but I’m able to provide context to AI assistant to ensure that it’s looking in the right places So you can see here beyond after the prompt library, there’s an icon for where you expect the AI assistant to answer from So is it referencing a particular documentation source? Is it the documentation for experience platform and CDP, CGA, Journey Optimizer that will tailor the answers for product knowledge? Or if I have access to multiple sandboxes, I can specify what sandbox I’m looking for answers and or performing actions for To ensure that it’s in the right place, even if I’m not in that application with the particular sandbox And then finally, the data view is related to and you’re seeing a lot of data views listed here because I’m in an internal org But that would pertain to any questions that or actions that you trigger from Data Insights Agent So for the purpose of today’s demo, the only one that matters is my sandbox Because all of my prompts are going to be related to audience agent and journey agent And then in the case of those agentic skills, we just need to make sure we’re referencing the sandbox where we want that information to be retrieved or for the actions to be taken So here I’m pointing to the LUMA sandbox and I’m going to start going into our use case So LUMA is a fictitious retailer that we use at Adobe when we’re demonstrating new capabilities And in this story, I work at LUMA, a retail company as an audience specialist I’m responsible for creating, managing and optimizing the audiences for my marketing organization And I’m working closely with Kira, I have access to real-time CDP And I’m working closely with Kira as my counterpart in the journey orchestration branch of my team To come up with a campaign that will be targeted towards our loyalty members Because we want to promote an upcoming promotion for our yoga apparel I should have said Pilates apparel to work with Alyssa’s certificate But a lot of people who do Pilates also do yoga, so I think she’ll be in our cohort Okay, so let’s start with our first prompt I’m going to keep it kind of general and say, what can you do? I just want to understand as I’m getting a lay of the land with AI assistant, what’s possible Now you can see it opened up a full screen conversation And in the AI systems response, it informs me that it’s powered by APs agent orchestrator There are a number of different tasks and or answers that it can retrieve A task it can perform or answers it can retrieve for me related to the agents across the applications And one thing to keep in mind is that this answer will change and evolve as we add more capabilities to the product So as we’re rolling out new skills across audience agent journey agent And the rest of the agents on agent orchestrator, you’ll see this answer change But within the answer, you can see what it’s able to do a couple of prompts that I can try And I can always reference back to the prompt library So I’m happy with this answer, I’m going to go ahead and give it a thumbs up You also have the ability to provide suggestions for improvements I’ll just show you what that looks like With the thumbs up, you can provide feedback on what worked well And with the thumbs down, you can provide feedback on either the visualization that you saw Maybe the answer looked incorrect Maybe there’s an issue that you want to troubleshoot All of that information is collected in triage Okay, so let’s start with my exploration for building an audience for Cura to leverage First, I’m going to invoke the field discovery skill to understand what fields I have related to loyalty membership So fields related to Those powerful fingers, there’s such a lot of typing when you did that Strong fingers from holding that balloon My other fun fact is that I can type 121 words per minute if I need to when under duress As fast as an agent Okay, so my first ask to AI assistant was I want to understand in this sandbox My Luma sandbox, what fields do I have related to loyalty membership Now in today’s world without an agentic skill This is usually kind of a back and forth that the audience specialist goes through To either work with their data engineer to understand what’s available in the XDM schema To leverage for loyalty membership Often times they also maintain an Excel spreadsheet outside of our system To know, okay, here are the fields that should be leveraged Even if we have multiple fields for loyalty membership Here are the ones that should be leveraged So we’re looking to capture all of that knowledge within this medium So that with just a simple prompt I can say show me any fields related to loyalty membership It scans my XDM schema within this sandbox and then returns an answer Now I want to go through a little bit of the anatomy of this response So you can understand what to expect every time So my initial prompts that show me fields related to loyalty membership And you can see reasoning complete So if we expand this window This shows the reasoning block that agent orchestrator used to interpret the prompt And then route it to the appropriate agent to either retrieve that information for me Or perform a task on my behalf And you’ll see examples of both in this demo So in this reasoning block you can see the agent orchestrator has interpreted my prompt So if we have a simple statement with even a lack of spacing It’s saying this user is asking for fields related to loyalty membership We will need to identify relevant data attributes And so it’s invoking the agent that can retrieve that information And you can see that it’s reasoned which fields within the XDM schema Feel them seem to be the most relevant for my objective Now within this answer you can see both a table that lists out all of the fields And their relevant score as well as the usage context for that field So if that field is already leveraged in an audience Or even already mapped to a member of an audience that’s mapped to a destination I should see that here in the card So I just expanded the window so I could see that a little bit better Shelby, in that example if I were the marketer I see Oh these two fields that look like they contain loyalty IDs Are they both like I would get a little confused like which one is the real one Is there any way to dig deeper into like maybe one has a million values And the other has five Is both real or is one the one that really has data in it Yes, so two part answer In some cases multiple fields will have values that are relevant So some of these could all be legitimate fields for audience building purposes I’m not going to rule that out It really depends on the context of your business And what you have ingested into your XDM schema But in this case if I wanted to interrogate a little bit further I can see that these fields are already a member of an audience So I could get a little bit more information So what audience is Hold on we got some typos What audiences use loyalty.tier So I’m interrogating that third field in the table to understand I see that the usage context includes an audience Let me see which audiences use it So now you can see the reasoning engine is now clearing operational data To understand okay here’s a list of all the audiences In this sandbox that are leveraging that field So this gives you indicator Daniel of the legitimacy of that field And or is it current right Is it being used in your newer audiences Is it something from before So you can kind of interrogate from different angles By the way I want to say I love your use of the word interrogate When talking about like clearing your data Makes me feel like I’m in like a cop show or something Yeah well The platform it says under the bright lights Or espionage spy operation that we set up This is great I can see what fields are available to me to represent loyalty If I wanted to build an audience I can also see what audiences already leverage That particular field I’m interested in But I can see you know we got some audiences related to yoga Some audiences related to loyalty members But let’s go ahead and build a new one And see what that looks like So let’s say I want to see take a workflow that I would normally do in the UI to build an audience Look sifting through the inventory to find the correct fields Dragging them onto the canvas and then estimating the size Let me try to do that via the agent all in one conversation So I really like that loyalty tier field It feels simple, straightforward So I’m going to tell AI assistant use the third field Audience of loyalty members who have been to my website Typos galore it can still interpret my ask And you can see the reasoning engine is calling audience agent now With its interpretation of my ask which is You had asked for the third field I have the context of the conversation within the session that we’re in So I know that you’re referring to the loyalty tier field But always check the agents work so that’s why I’m interrogating it And we are going to also include the added criteria of loyalty members who have visited the website Let’s see I’m going to try the prompt one more time because my expectation is that it should have invoked what’s called a plan based response And first outline the steps in the plan to create that audience So I’m going to give it a chance While there’s queries, Rob in the chat had a question about how the agent determined which fields represent loyalty Great question So while let me see if I can scroll up while it’s thinking hopefully that Let’s see You go back here So you can see I want to show you the happy path So if it’s thinking too long, that’s too long for me Okay, so going back to the question in the chat How did it identify these fields? Each response for field discovery does include One of the results So you understand from the agent’s perspective why it chose Why it chose the fields that it chose in your exeens schema And also why it marked it as anywhere from highly relevant to just To the next page here And that’s as well Okay, so if we move forward into the audience creation Let’s see We’ve completely lost our plan Let me try this again One second, bear with me The live demos always get me Now the good news is we’re calling audience agent And what I’m waiting for is for it to invoke a plan So that we can see the steps that the audience agent will take To identify the criteria for this audience accurately based off my exeens schema And then estimate the size of it Let’s see if we can proceed Okay I’m going to try one more thing See if I can get back to a conversation Okay, so this can show you the audience creation plan Now, in the happy path world where everything is firing on all cylinders My expectation when I had asked to create that audience Is that it would invoke audience agent and then kick off the audience creation plan So this is what’s known as a plan based response from agent orchestrator And it outlines the multiple steps that the agent will take on your behalf With your approval Overview and approval to proceed through the steps of the plan And ultimately perform the action So anything that generates an action is likely going to invoke a plan Rather than being a one shot So the first portion of the plan first interprets the goal Okay, this user wants to create an audience of profiles for loyalty status Who have also visited the website The first step is to identify the audience characteristics and define a definition The second will be to estimate the size of that audience The same as what you would be doing in the rule builder in RT-CDP And then finally drafting a proposal of what that audience will The final audience will look like and then saving it on your behalf with your approval I’m going to try to move forward To the next step We will be activating audience Where that’s coming from Okay, so you can see it’s proceeded to the second step Which is estimating the size of this audience You’re presented with the estimated profile count And a description of what it’s estimating As well as the profile query expression So that you can see, okay, this is what the definition would be In the eyes of the rule builder So let’s go ahead and save this So that Kira can get going with her journey Save this as So I’m going to specify the name And one thing to keep in mind is as I’m saving this audience It will, once I approve the proposal And allow the agent to save it This will save it in RT-CDP And it will be available to Kira to leverage in her draft journey So I’m submitting that We submitted the decision The agent will only perform an action on your behalf if you approve it So that’s why it’s important to review the proposal And then approve the proposal so it can go ahead and create it You can see here it’s hyperlinked the audience for me And I can find it directly within RT-CDP Now one point before we switch over to Kira Is that this is available immediately for Kira to leverage In the journey that she’s going to draft using journey agent If we wanted to do this flow end to end within the conversation Meaning, now I’m ready to draft a journey with journey agent I’m going to retrieve this audience that Shelby just created Today there’s a 24 hour delay between creating the audience via the agent And then having the agent be able to pick that up again via the conversation And that’s just due to the latency of what’s available in our knowledge graph Which is what AI assistant leverages for all of the information retrieval But for practically speaking If we were looking to do a handoff in real time Where I’ve completed the audience creation Now Kira can pick up where I left off with drafting the journey This audience is immediately available to her in the UI And so we’ll showcase that in the demo But there you have it So we’ve used audience agent to both search for fields related to loyalty membership And then identify an audience criteria on my behalf Estimate the size of the audience and create it with my approval along the way But with just a couple of simple prompts we created an audience Back to the results I wanted to say to you, well, while Kira picks up, shows her screen I think what’s really important in all the steps that you’ll see with our agents Is that we’re very transparent on how the agent comes to that answer Or how it comes to that conclusion Like it’s really important for users to always go back and validate the agent’s work It’s not fully replacing your workflow today It’s just allowing you to have a little bit more visibility on the steps To all these different jobs that the agents are doing So I think we’ve gotten a lot of feedback from our customers At least that they appreciate the agent actually showing how reasoning works And showing how they got to an answer Absolutely There are a few questions that came in I wanted to check because I think this slide said that the audience creation was coming soon But Shelby just demoed it For customers who do have this capability Can they create audiences right now? So we wanted to show this sneak peek for everyone because we’re very excited about it This is a capability that we’re working on And you should be able to see that soon We’re aiming hopefully within the next couple months And so you’ll be able to see that soon if you have access to our agents You’d still have to work with your Adobe representative To understand how you can get access to our agent capabilities Okay And Rob was following up on his question about how it knew the loyalty fields Does it just look for the term that Because we use the word loyalty when we’re looking for it Does it also use synonyms for loyalty Or can it infer things like somebody who’s made repeated purchases Could it infer that that is a loyal customer? Mm-hmm Oh I love that call out So today in the response that you saw It’s fairly rudimentary It’s looking at the name and the description And then the potential associated values with that field So for example loyalty tier If I can see that the values are platinum, gold, and silver We can deduce that that is likely related to membership status But we have plans on the roadmap To start adding a more sophisticated semantic understanding of the XDM schema To make those types of associations that Rob is calling out Where even if it doesn’t say on the face of it loyalty tier We can deduce that the profiles associated with that field Have performed high propensity activities in the past And or they are logged in in some capacity And have signed up for our membership program So that’s coming but thank you for the question Great cool and I know there are more questions We’ll come back to some at the end Or if not address them in the community thread That will be posted for this episode But let’s get to Kira so she can show us the journey agent Alright thank you Shelby for the demo I’m excited that we now have an audience we can use I’m going to showcase two journey agent skills First it’s going to be looking at a fallout analysis of a previous journey that I’ve done And then we’ll look at a preview of the create As Shelby showed with audience agent We have the same thing we can preview with journey agent So in this case again Shelby mentioned We are working for a brand called Luma They are a wellness company And we are going to advertise some of our special classes to our loyalty members So I’m going to create a journey of promoting a goat yoga class that we’re doing Previously though we’ve done a puppy yoga class And so we had a journey to advertise that So before I get started in my goat yoga promotion I want to look back at my puppy yoga promotion and see how things went there So there’s several different things I could analyze here But I’m going to ask the assistant here to analyze fallout for my puppy yoga journey And again I don’t have to have the specific name of the puppy yoga journey here It will infer this But if I want to get the exact one I can hit the plus button here and then type puppy yoga And it will search through all of the objects that we’ve got And then the assistant knows exactly what to reference So if it is inferring something that you don’t like You always have the option to actually get the exact object you’re looking for So I’m going to press go here And then you hit it again Go ahead Kara you hit the plus sign to bring up that menu? So you just hit the plus sign And then all of the objects that the agent can reference will show up here And it will filter down based on what you type in it So if I type P-U-P-P-Y This is probably the only thing close enough to puppy I don’t think there’s a lot of objects named puppy in this instance other than this journey So it found it very quickly But we could do any other one too So if I type yoga You can see I’ve done some different promotions previously for journeys Regarding yoga gear So it would help me narrow down the objects I’m looking for Nice So this has come back And again as Shelby showed It shows the reasoning that it went through to achieve this answer But what it’s telling me here is that my luma puppy yoga journey Is that there’s a critical issue at an optimized node So it says there’s almost 60% of users dropping off And that’s about 4500 users that are not getting what they should be getting So it’s saying also that my silver and bronze loyalty levels Or communications aren’t really happening because no users are progressing from the condition to those emails that I had previously set And it actually provides me a look at what the journey looks like and where the problem is occurring So it says I’ve got my audience here And then this optimized node is what the problem is here So it’s saying that 60% of my users are falling out When I wanted them to go and receive a different email depending on their loyalty tier So it looks like my platinum and gold members here are working But then my silver and bronze tier members aren’t really flowing through So that’s something I definitely want to keep in mind for future communications And we’ll be sure to make sure the AI assistant gets the data conditions correct in this next one For my goat yoga Before I get to goat yoga, so we know the agent will always provide you a recommendation on what you can do to fix those things that it identifies as an issue So here’s an issue, it’s giving you a recommendation to go in and fix whatever is wrong I’m not going to focus on that today because I’m going to just continue to build a new journey And we will just get started with that So this prompt is kind of long but I’m going to paste it in So rather than doing just an email promotion this time for our goat yoga class We’ve decided hey let’s expand and see if we can target on other channels too And so I am asking the agent to create a journey targeting loyalty users I’m going to ask it to check and see if there’s a valid email address And if the profile has an email address, send an email with a promotion for the new goat yoga class And also include a little promotion in there If the user doesn’t have a valid email, we want to check and see if they have a registered phone number And if they have a phone number on file, we’ll send them an SMS message So then we’re going to wait four days and send a follow up nudge if they don’t open any of those things So I’m going to hit go and the journey agent is going to analyze my prompt And then come up with a plan on how to tackle it And one thing I will call out as well is as Shelby mentioned What I put in here was targeting loyalty users I could have called out the exact audience if I would have wanted to But those need to be, they sometimes have a 24 hour delay So I’m going to actually go into the UI and pick the audience that Shelby created But for now I’m going to have a placeholder audience here But as you can see, the AI assistant brought up a plan here on how it’s going to help me create this journey So it’s saying let’s start by collecting information about journey creation It’s walking through my plan and it’s asking if that’s a good plan to proceed with and I will say yes One thing you’ll notice here as well is there are two different things the agent will do So it’s looking to either, right now it’s trying to decide what kind of journey we’re creating And then it will do either step two or step three because those two journeys are different So if it would be a triggering event journey it would know that and go and find an event that would work But in this case it should see that this is a target audience for my journey So it should search audiences and you can see that was just crossed off because the agent detected that this is a read audience member, journey member Alright, so now we’ve got the prompt back that says hey, we know you’re doing a read audience journey Do you want to proceed and find the target audience? I’m going to say yes And then it’s going to look similarly to how Shelby was looking for different fields It’s going to look for different audiences that could be related to loyalty Again, as you create your objects in Experience Cloud We really highly recommend that you include as much metadata as you can in the description of the object or in the title of the object If you would like to use the agent because it will definitely improve your experience Here it’s come back with some audiences that I can do Again, Shelby’s isn’t available here because she just created it But I’m going to pick a placeholder audience and I will just pick number 5 You don’t have to type in the full name of the audience here, you can actually just select a number that it put back And then with that it will go and create a draft journey for us And I think you answered one of Rob’s follow up questions which was about whether or not the prompts could see the descriptions of things like the fields and the journeys And it sounds like it can Yep, it definitely does. So we highly recommend including a description there And you’ll see in this step 5 when I get down here is that the agent actually helps me save my journey that I’m creating with some metadata added in automatically I can of course adjust and add additional metadata But it will also help me get the right information into the description and the title so that the agent in the future can come back and actually look at the correct objects So here it says, hey I’ve successfully created a journey and personalized it One call out here is that we won’t have the content in the actual emails and SMS created from the agent just yet But we have the flow of the journey created for you. So it gives us an overview of what it’s done And then it asks, hey do you want to proceed to phase 5 to save the journey as a draft Before we do that, I’m going to just close this so you can see a little bit more I can preview the journey that it just created So it chose that audience that I asked it to It’s checking for an email If they have an email it’s sending an email promotion No email, it’s checking for a phone number And then sending an SMS Then of course it’s going to wait four days and then send a follow up nudge That’s exactly what it wanted. So that’s great I’m going to hit the thumbs up here And it was correct So I will press submit And then I’m going to answer the question that said, do you want to proceed to phase 5 And I will say yes And it will actually help me save the journey as a draft And I’ll show you what I mean by including some of the information in the metadata description So it says let’s review And it auto populates this description here with some of the most important pieces So I will save that And then it will hyperlink me the journey that it created And I can go in and add the audience that Shelby created previously I like that it does the description I think that’s something that a lot of us tend to skip over when creating things Just because we want to get things done So the fact that it’s writing a very detailed description is a nice time saver Absolutely, absolutely yes So you can see here I just opened my journey and I went to that audience node And I can on the same day select the audience that Shelby just created And now my journey will be ready with a few touches of review to activate So that is our journey agent capabilities Some are upcoming, some are live now We’re really excited for you to use them That’s great Kira There was a specific question about journey agent that came in early in the episode That was from Deepak asking Can we identify which journeys are currently using a specific audience as a trigger? Yep, that is an available feature We didn’t show it today, but you can do that with the AI assistant You can essentially ask it to analyze audience conflicts or schedule conflicts So if you were to type in, hey, analyze my conflicts It will look at both audiences and schedules amongst all of your journeys And see which ones are using the same things And I guess that was also an audience agent question too For maybe if you had an audience definition that included sub audiences Shelby, could you use audience agent to determine which audiences use a specific audience in that audience definition? Which audiences use a specific audience in that audience definition? So from the field discovery direction that I showed in the demo If you see usage context for audience That badge that said audience That does account for segment of segment Meaning that field is a member of an audience that’s a member of an audience So that’s acknowledged from that direction It sounds like the question is also if I was looking at the definition of an audience Could I see that an audience is referenced? Is that the question? It seemed like what it was getting at Yeah, that should also be possible just by leveraging operational insights through AI assistant So understanding the composition of an audience What attributes or events make up the definition And if that attribute includes a segment membership then that should also be called out One thing to call out as well for audience create as a skill is that we are soon going to be rolling out support to create an audience off of an existing audience So if you’d already created a seed audience and you wanted to expand that definition or leverage it to repurpose for a new campaign and a new audience You’ll be able to do that via the audience agent very soon Great So I know there are a lot more questions in the chat What we’ll do is we’ll create a post in the Experience Lead community for today’s episode And any of the questions that we didn’t get to we will respond in there to hopefully answer all of your questions about these cool new features So with that, the way we like to end every episode is with an unrelated cool tip And today’s unrelated cool tip is brought to us by the Product Marketing Organization And Alyssa, what is the unrelated cool tip you would like to share with our audience? I learned this from my mom I like to dampen like a paper towel and put on top of my plate of food when I microwave it So that my leftovers aren’t dry the next day So if you have leftover chicken, you don’t want dried chicken when you microwave it, dampen a paper towel, the moisture within the paper towel on the plate will allow the chicken to prevent it from getting dry in the microwave or pasta or anything else that you don’t want to be dry It works like a child Well, it’s lunchtime here on the East Coast, so we’re going to try that trick out Thank you so much Shelby, Kira, Alyssa, this was a great episode and we look forward to you having you back on the show to show us some new, even newer cool things that you’re working on Thank you so much Thanks everyone