51ºÚÁϲ»´òìÈ

Transcript

One. Thank you guys for joining us today.

For today’s session. For today’s tech sessions, I’m super excited to be presenting, with Rachel and Ariel. We’ll get started soon. But before we do a few housekeeping notes. Throughout the presentation, there is a Q&A chat. So we strongly encourage you to ask any questions that come to mind. However, rest assured that the last five minutes or so of the presentation will just be dedicated to asking and answering any of your additional questions. Lastly, a link to this recording will be emailed to all of you in about 24 hours, and then another recording will be posted into a screen for you to watch. Thank you guys and I’ll hand it over to our presenters. Now.

Thank you so much. Can you have help switch to actually presentation mode? Thank you so much.

There we go. Let’s start from the beginning.

All right. Thanks everyone so much for joining. Rachel and I, we’re stoked to present to you and jump start your day or your afternoon evening on getting, more productive with generative AI and best practices with our in-house AI assistant tool for Real-Time CDP, Journey Optimizer and Journey Analytics. Customer journey analytics. Rachel has a wonderful background and and generative AI as our product manager, connoisseur. And she is leading our AI assistants for AP. Lots of product, and product knowledge and expertise for 51ºÚÁϲ»´òìÈ Experience platform. And she’s really excited to bring smiles to our customers. Particularly fond of tennis and volleyball. I also used to be a tennis player, so I can relate. And I my background is I’m on the Journey Optimizer team and also in product marketing for a API assistant. My background is in video production and content marketing, and I’m really excited to work with all the customers to bring more product value to you.

So today’s agenda, we’re going to go over kind of the state of AI and a bit of overview on AI assistant. We’ll give you some example prompts. Rachel will do a wonderful demo for us, and we’ll talk about how AI assistant is built with trust and data usage around that, as well as how can you can access. We’ll wrap it up and provide give you time for some questions for us to answer.

So let’s get started with the the landscape of AI. Generative AI. The adoption wave has been extremely transformational. It’s one of the fastest adoption waves we’ve actually seen in tech history. As many of you are aware, and it’s been anticipated to evolve ever so rapidly. Marketers are actually proving to be extremely strong advocates for adopting AI tools to drive better personalization across their customer journeys, and be more effective and efficient in those ways. In fact, 89% of marketers and customer experience leaders believe generative AI will help them with better personalized customer experiences. 42% of generative AI survey responders are anticipating marketing applications as having the most promise for their organization, and even in the next two years, AI driven features will be embedded across business, technology and categories, as well as 60% of organizations will actively use such features to drive better outcomes without relying on technical AI talent. So there’s a lot of movement we’re seeing in AI, generative AI, and we’re really excited that we get to participate in this wave. So I assistant for, 51ºÚÁϲ»´òìÈ Experience platform, it’s that simple conversational interface powder powered by generative AI that will enable you to ask questions and streamline workflows for that enhanced productivity. So where it’s built for, as I said earlier, Real-Time CDP, Journey Optimizer, and Customer Journey Analytics. So a bit about how this works and our capabilities for AI system. Right now it’s generally available the product knowledge part of AI assistant. So you can quickly get access and understand, product knowledge like concepts and learning and troubleshooting for all those three, tool, products that we have within the 51ºÚÁϲ»´òìÈ Experience platform and within public beta, we have, it’s called Operational Insights. And so you can learn about your business objects. This is audiences, schemas, data sets, destinations, journeys and this will help you understand the lineage, the usage and the hygiene. This is specifically for Real-Time CDP and Journey Optimizer. So it’s a really helpful tool to help you deepen your knowledge, improve that product, product proficiency, increase your streamline operational workflows.

So we have three identified personas that are best leveraging the AI system tool, from data and AI teams, we’re trying to best understand data pipelines and how it’s stood up and maintained. And this is to help, accelerate that data exploration, data management, data insights. And then we have marketing ops teams who are responsible for managing and tracking performance of marketing workflows and the usage of those business objects like audiences profiles. Then we have our marketing teams. They want like a self-serve of, to leverage like product functionality. How how can they best learn how to use the products that are, within the API? And what are the objects that can most efficiently, execute better marketing campaigns and IDs faster and better? So there’s these three personas, and we’re seeing them all use AI system.

So we want to get into a bit of, example prompts across the two parts of the AI system tool.

Having issues advancing actually.

I’m having some issues actually advancing the deck. So now we’re on the oh yeah, there we go. Okay, my deck is totally frozen. So, are we actually on the example prompts. Example problems for product knowledge. Wonderful. Okay. So for product knowledge, what we have here are, pointed questions you can asked. So it’s like for learning, it’s for open discovery and for troubleshooting. So some examples across the three products is like what is the difference between a campaign and a journey around open discovery. You can ask can I run an email experiment. And then for troubleshooting it’s been very helpful for the different personas. I explained, why, how long does it take for data to come into customer journey? Analytics? So my experience frozen. You can help me. Let me know if I’m actually on the next slide. There I go. Thank you. So for operational insights, some sample questions. To kind of get you started here are around these three categories. So that could be usage lineage or data hygiene. So really nice example prompt features. Which segments have never been activated to destinations or journeys. That question can be for both products. For lineage that would be do I have any segments that are used in other segments? And then for data hygiene, do I have any journeys that have never been used or have tests in the name? So that’s all really helpful prompts. You can start to get those gears turning and how you can use AI assistant. So I’m gonna switch it off to Rachel, and she’ll dive a little bit deeper into how how the tool works.

All right. Thank you so much, Aria, for that great overview of AI assistant and digging into some of the use cases. It’s great to meet you all today. So my name is Rachel Hennessy, and I’m the lead product manager on AI assistant in AP, and I’m excited to get into a demo for all of you in just a few seconds. But first, I want to encourage everyone to use the chat pod. So if you have any and every question, please pop it in there. Questions about how to get access, questions about use cases, questions about upcoming roadmap. This is your session. We really want to take and answer your questions. So with that I will share my screen.

And get this demo rolling. And especially as even as I’m demoing, please feel free to ask questions. Live and we can kind of get into it while I’m screen sharing. So hopefully you can all now see my screen of 51ºÚÁϲ»´òìÈ Experience platform. And so for this demo, let’s say that I’m a marketing ops specialist and I’m trying to clean up loyalty audiences. So first let’s just start you know, let’s say I’m a new user to AI assistant. And I want to understand what can I help with. We actually just integrated with Unified Search, which is this toolbar here. So I can ask what can I assistant help with. You can see that now in Unified Search we have some options including ask AI assistant. So I’ll go ahead and click that. And that opens up the right rail which is the persistent AI assistant. And it’s asking I ask this in the question. So it’s letting me know that AI assistant is designed to help users in a variety of ways. Talks about learning, troubleshooting. These are a lot of the use cases that RL just went over and gave some sample prompts. But here, if you ever forget or want some guidance on what AI assistant can help with, feel free to just ask AI assistant directly because that’s what it’s there for.

You’ll also see the sources that this came from. So whenever we answer informational questions, AI assistant is also going to give you the sources that it was grounded in when giving you that answer. And you can feel free to just select one of those links and it will bring you over to the resource that it got that information from. And so today I assistant pulls from experience league documentation. But we just pulled into beta actually tutorial guides, product legal documentation and even knowledge help support articles. And we’re looking to expand further to community forum information and more. So you’ll just see the, kind of brain power of AI assistant continue to increase some other features to call out here on this more product knowledge related, question. You can see that there’s actually inline citations that show you exactly where this certain piece of information came from. So I can select this three and it pulls me directly. So like we said, you know, use AI system to find supported, experienced platform objects. We selected the three to learn a bit more. And now it’s pulling me into this question guide for AI assistant, where it’s giving me sample questions on different use cases.

Another place where you can find sample questions are actually behind this light bulb icon, which we call discoverability. So when you select that, you’ll actually be able to see a full list of what AI assistant can help with in this product context. So I can select this here and actually get some additional details about what types of questions I can, ask AI assistant and audiences. This is really important to understand because, you know, a AI system does a clone what we call a closed domain chat bot, meaning it has a specific scope of things it can help you with. It’s not going to help give you suggestions on, you know, what kind of sandwich to make for lunch. It’ll be very specific in terms of what it can help with on 51ºÚÁϲ»´òìÈ Experience platform. So it’s great if you come here and kind of get some inspiration about, what you can ask our assistant today. These are heart hardcoded meaning. They’re like a preset list of questions. But in a couple of months, we will be dynamically generating these prompt suggestions. So you’ll see many new prompt suggestions, every time you come here.

All right, so back to my purpose. Here is the marketing ops, specialist who is trying to clean up loyalty audiences. Now that I have a pretty good understanding of what I assistant can be used for, and I know where the resources are to learn more information. Starting with just asking AI assistant itself. I can start to ask some questions, so I’ll start asking list the audiences is containing loyalty in the name which were created in the last quarter.

All right. So I’m asking AI system to question and natural language. And just to give you a little, understanding of what’s happening behind the scenes, we’re taking this natural language question. We’re translating it into a SQL statement and then querying a database called a knowledge graph that we have actually set up behind the scenes of every single customer sandbox. So here we’re finding, okay, there’s a segment ID with this specific name, Luma loyalty with gold or platinum that was created in the last quarter. So now I have my answer that there was one audience created with those constraints. I’d also like to point out that AI assistant goes through and explains exactly how it came to that result. So it’s giving this kind of step by step first to considered all segments. Then it just took the ones with loyalty that it created or took just the ones that were created in the last quarter. So this is basically a natural language explanation of the SQL that was created. And then for the more technical among us, you can go to the source and actually view the query and see the exact SQL. And the reason we’re showing you this within the product is to make sure that you can trust the results. So we want to make sure that one, you know, that AI assistant interpreted your question correctly. And then to that it actually, you know, ran what you were wanting to ask so that the result matches with the information that you’re trying to receive.

Great. So we just found one audience that has loyalty in the name, but they don’t want to see all the audiences. Right? I’m a marketing specialist. I’m trying to clean things up a bit. So list all the audiences containing loyalty and the name and show how show how many profiles they each have. Okay, so now I’m kind of expanding my search. I want to see all of the audiences with loyalty and the name, and I asked for some additional information, which is how many profiles they each have. So I said changes. Again, converting my natural language query into a SQL statement and looking, in the knowledge graph that’s specific to the sandbox. So here I have my answer. So I can see the name of these five audiences with loyalty and the name the associated profile account. And I can download as a CSV. So this was is kind of a newer functionality that we have, which, you know, in the past we would actually limit the results to just 25 rows. Now we still limit to 25 rows, but we allow you to download CSV so you can get thousands of rows offline to manipulate the data as you would like.

Again, we have that same kind of explainability where we’re breaking it down. I’d also like to call it that. Each of these entities are hyperlinks. So one things that, customers have really enjoyed about this as being able to just, you know, select the audience from their and be navigated directly to it. So no more kind of searching and trying to find it. You can just simply, select it from the chart and it’ll bring you exactly where you’re going. All right. So let’s see that now, I actually have to start a new conversation, so I’m just popping that one back in because I want to show, I the ability to ask follow up questions or your ability to ask follow up questions based on the previous prompt. So once I have my list of audiences with loyalty in the name, I want to go a step further and try to understand where are these audiences used? Are any of the above audiences used in other audiences? So I know that many, of, implementations that you all have have audiences that kind of reference other audiences and that can get pretty complex and hard to figure out where everything is used. So we’ve built that into assistant, where we’re able to ask those sorts of questions. And here we have assistant letting us know that the audience Celtics bandwidth Boston and gold loyalty is used in another audience.

So maybe now I want to dig a little bit deeper into this audience and understand what are the actual booths that are used in. And now I want to showcase, that act or the object autocomplete functionality. So you see, I just hit the plus sign there. I could have also hit the plus sign here and started to type. So what this is going to do is connect my query directly to that object, making sure that, we’re knowing what what they’re talking about here. So audience and it also stays on typing. So you don’t have any typos or have to worry about typing out a whole name of something because, we’ve seen some of these names get get pretty long. So here I’m searching for the attributes that are used in this audience that in the past, I’d have to kind of open up the audience and find that here. Just print out the list very simply. Maybe I want to go a step further in the lineage and figure out where do these attributes exist, what schemas are they part of, what schemas I’m taking these attributes so you can see I’m kind of doing the lineage down from audiences used in other audiences. What attributes make up those audiences? Okay, where are those used in the schema. And you’re able to do all of this in, you know, just a few seconds with, with AI assistant, instead of having to write up the really complex SQL queries or go to that one person on your team, who knows how to figure all of this out? It’s it’s making it much more accessible to everyone. Cool. So here I can see the list of different schemas that contain those attributes that that’s a lot of schemas that I would have had to search through manually one by one, to see if those attributes existed.

Great. So let’s see lastly here. So now I am noticing that there are quite a few schemas that contain those attributes. I might be wondering at this point, is that okay, like are there best practices for schema creation where maybe I shouldn’t have attributes shared across multiple schemas? So I could ask, I said, what are some best practices for schema creation? Is it okay that I have attributes that are used across many schemas? I can probably handle that, but I’ll make it easier. So now we’re back to more of a product knowledge question asking for, some information for my assistant. And so here able to give me a question in just a few seconds, saving me a lot of time from going to experience, like trying to figure out these best practices, getting an answer to my question of using attributes across many schemas. In the past, I may have had to bug my data architect who is specializing in this, but now I can just take it into my own hands a bit more.

I can also continue my learning journey using these related suggestions. So how can I ensure that attributes are consistent? I can just fire that question off. And again, it’s always going to show me the exact, document it came from. You know, it might have taken me a really long time to find this. Best practices for data modeling tips. Now I just find it, you know, ask a question. You get that resource right away and I can bookmark it for later use.

So that’s all that I would like to show in the demo today. We have a ton of new capabilities and features coming on the AI assistant for AP roadmap, which, we would love to to share with you. I’ll give you a little bit of a teaser into that, and at the end of the session, but any I do see that there are some questions in the chat. I don’t know, if there are any that we should take now or hold for later, I have that, yeah. There’s actually, like a few questions that would be helpful if you can answer now. One them be check was saying what did you click on to get the unified search bar. Yeah. So if we go back to the homepage.

It’s right on this, home page. It’s just right. Right in the center there. Yeah. That’s really helpful. Thank you. Yeah. And then the other questions, I think we can. Okay. This one’s another one, too. Might be helpful. Are the objects available to reference in the question limited to the current QA Q&A session with AI assistant? Or are objects not referenced in the conversations also available to use? Hold on, let me see. Are the objects available to reference in the question? Yeah. No, I’m not sure how to answer that question. I think I might understand, so maybe I’ll take a step back at the objects that you can ask about in AI assistant. Are any of the objects that are in this sandbox? So for example, in this demo environment it would be the objects that are in prod V7 in, what? And more specifically, to get that list, I would recommend coming to this discoverability panel or it’s also an experience league. But sometimes we find it’s easier for folks to just go and product. So audiences attributes, data flows, data sets, destinations, journey schema sources. These are the objects that you can ask about, the limited QA session.

And I’m not sure if I hope that answered your question. And then there’s one more that I think would be relevant to answer now too, is, yeah. Can we use AI assistant for QA and insights in anomalies in audiences, for example, we can see a big difference on audience X for day Y in the last month. Great question. So that is very top of mind. We actually, just pushed to our production environment some of those functionalities yesterday. We have a feedback program group that we, meet with every week to test out some of these new functionalities so I can, maybe we can work with you on that. Get names of folks that might be interested in participating in that, but some of the new capabilities that are coming for audiences are significant change detection. So kind of similar to what you’re talking about with anomalies, audience estimation. So very simple estimating sizes of audiences based on attributes and then propensity to propensities and churn. So more on the prediction side. So those are the three use cases that we just finished building. They’ll be entering our alpha beta feedback program. Next, next Thursday we’ll release them. So if you’re interested in participating in that program on a weekly basis, we’d love to, get you involved because. Yeah, I assistant, what I just showed this is just the start of it. We have, a long list of other things that we’re going to help it, work. For, right. So far, good feedback on answering these questions. I just have one more for you, and then we can just move on and answer the rest at the end.

So the question for Marianne is, how can I get a list of audiences that use this specific event value or attribute? Yeah. So that is something that you can do today with AI assistant. So for example, I showed, you know, how many audiences have loyalty in the name. I could also ask how many audiences have and then put in the specific notation of a field. So like, you know, how many audience use percent name dot last name. And it would come back with that. Today you do have to be a little specific in giving the exact field path. However, another new functionality that’s coming soon is, called Xdm Field Discovery for AI assistant, and that will make it even more powerful. So you could say things like, show me all the fields that have information on loyalty, for example. So not just searching specifically on the loyalty name, but anything that has that similar semantic or show me all the fields with that last name, or audiences with the last name field. And then lots of different, you know, we know that you have lots of different fields that might be named slightly differently depending on what data set they’re connected to. So hopefully that answers your question that. Thanks, Rachel. Yeah. Let’s continue on and then I’ll answer the rest of the questions at the end. Cool. Sounds good. I am going to stop sharing this screen, and then we’ll pop back up the presentation.

All right. Fantastic. So we went a little bit over how to find example prompts. But just want to drive this point home that it’s really important for you to, to take a look at these sample prompts. So important that we actually are two of our, features that we’re really working to build, and we’ll have them in the product by early next year for the general availability, our prompt autocomplete and related suggestions for operational insights. So we’re sure we’re giving you some examples of questions that will work for you. So that’s something that we’re really committed to doing. But in the meantime please check out that discoverability panel. That’ll be your best bet for learning what types of questions can be asked and answered. And then also just just try it out. So this is completely free for all customers to use. It’s, you know, unlimited consumption for you. So please just test it out, play around with it. Have fun with it. Give us feedback in the product by giving us thumbs up, thumbs down. We look at all of that. We have a whole annotation team that’s looking through to try to improve the experience for you all. So we’d love to kind of be your partner in making this better for you. All right. So trust and data usage. So when we think about trust considerations for AI assistant. And so by now I’ve mentioned a few times that our alpha beta feedback program, we spent a lot of time working very closely with customers to make sure that, we build this AI with what is expected, from all of our customers. And so we’ve implemented all of the controls that have been, desired.

One point to, call out is that it’s not, by default. So it’s off by default for all of our users. But then we allow admins to add granular permissions by product and by capability. So the admin can say if their users have access to it, new CDP new, and if they have access to it and Customer Journey Analytics. And then also if they have access to that documentation type use case, which we call product knowledge. So asking questions and getting answers from experience league. And then separately they can choose do the users get access to the operational insights. So the questions about, you know, lineage and searching for objects all access controls are protected. And so what I mean by that is the AI assistant supports role based access control, object based access control and attribute based access control. So to give an example there, if aerial does not have access to viewing audiences, if she goes into AI assistant and says, show me all the loyalty, audiences AI assistant will say, wait there, you don’t have access to audiences. We are not going to show you that. Or if I ask for all the fields in a certain schema and a couple of them are actually sensitive and I don’t have access to those. If I say I said show me all the field in the schema, it’s going to pull out the fields that I don’t have access to. So it’s completely, it follows all the access controls. So it has gone through all of the HIPAA review and is a HIPAA ready feature as.

Some data usage principles. Purview. Because I know this is very top of mind. AI assistant does not even look at the end consumer data. So we got a lot of questions. You know, are we using the data to train any models? AI system is not even looking at the data in your sandbox at this point. It’s looking at the objects, right? You saw me ask questions about audiences attributes, not about the data within those fields or within the profiles.

We do look at user prompts to improve the product. So. Right. I mentioned that it’s really important that you give us thumbs up, thumbs down, and feedback about how it’s going.

If we didn’t get that feedback from you, this would be a pretty stale product that, you know, we wouldn’t be able to improve. So it’s really important to us that we looked at the prompts that come in, look at the responses that we’re generating, make sure that we’re, you know, measuring any, any issues that come up, make sure that we understand, you know, what are the important use cases to solve for.

And then we are using Azure Open AI for parts of the processing. Open AI does not use any data sent to it to improve. And the logging is disabled on there. And then we do use open AI’s content filtering as well as our own homegrown PII filtering.

And so we’ll only respond if prompts pass those guardrails. And then last, each answer, as you saw in the demo is verifiable. It has citations, explanations step by step. Explanation of how it came to the answer. And we also share the SQL query.

All right. Getting access. And I think skimming through the, the chat, I did see a couple of questions on getting access. So hopefully this is top of mind. There are three steps to getting access. The first is organizational access. So I think the question that I saw is just coming through the chat was, we see that we see a pop up when we click on the chat icon that says, hey, you need to, your organization needs to fill out some legal papers that is this first step, organizational access, I think. I don’t know if we can send a link, right now to this group, but we’ll certainly send it after there are some general artificial intelligence feature terms that you all can, have the link to, and you can work with your account team to make sure that that gets, signed for your company. That will be the first step to unlock access. The second step to unlocking access is that permission step that I was talking about, how the ad administrator needs to give access on the product level and then on the capability level. And then the third step, once you’re sorted from a company level, sorted from a permission level, when you click on the chat icon, you’ll just have to agree to some user terms that you can read through, and you’ll accept. And then you can start using the feature.

Hopefully that answers some questions there, just to give more of a deep dive into what that looks like. So the permission screen for our for real time customer data platform and 51ºÚÁϲ»´òìÈ Journey Optimizer is going to be in the permissions UI. So there you can add the it’s called AI assistant permission. One of the permissions is enable AI assistant. So that will open the chat rail, give you the product knowledge capability. And then there’s another permission for View Operational Insights. And so as we continue to develop capabilities we’ll give more granular permissions so that you really have that flexibility to allow this group of people to that capability. Maybe this other group of people, different capabilities.

And then in the admin console for Customer Journey Analytics, that’s where you’ll turn on product knowledge for Customer Journey Analytics. Customer journey analytics also has a data analysis use case that they are in alpha or in two weeks will start their alpha program there. So that’s a very exciting one coming as well.

All right. To wrap up, I know I’ve kind of nodded to a couple of the things that are coming in the pipeline, so I’ll give a little bit more context on what those are. So if you look at these five, sorry, six rows of what I assistant can help with, today we’re more in the information retrieval and content generation phase. So we showed examples of product knowledge where we can get quick information on specific pointed questions or learning, help with understanding the features. So open discovery early and then also with troubleshooting and best practices we also provide information retrieval on the object level. How many audiences do I have? How many of these audiences have never been activated? What am I most used? Most used attributes. Are there any attributes that have never been used downstream? So those type of information retrieval use cases, we also have content accelerator for 51ºÚÁϲ»´òìÈ Journey Optimizer which is an AI assistant functionality. So with content accelerator you can generate copy and images that are brand compliant based on a brand kit that you upload. So those are the two use cases that are generally available today. And then we have a bunch more in the pipeline. So for data analysis and interpretation, we’re enabling through AI assistant the ability to ask about charts and generate graphs, summarize reports and get insights through customer support. The next use cases coming are creating a support ticket within AI assistant. So we know that when you go and create a support ticket, now you have to kind of add a lot of information about what’s going on. Our vision for customer support is that you start with asking AI assistant some questions, and if you’re not able to solve your issue through assistant, you can just file the support ticket right there and it will have all the context from your discussion. Additionally, with customer support, we’re looking at some more diagnostic type workflows where AI system will be able to figure out the issue at hand on the optimization track. That’s where those some of those audience and journey use cases come in. So earlier I spoke about the audience recommendation, anomaly detection and the prediction and estimation use cases. So those are being actively developed or tested rather now and then. Last is around workflow automation. So this is around creation of audiences and journeys.

Maybe, you know, if you want that we’ve heard the use case from some of our users that it becomes really tedious sometimes if they have to take an audience at change one attribute, but, you know, create like 20 copies of them. Those are the types of things that we want to use AI assistant to automate for you. And so we believe that a mix of these will be used together to accomplish, a certain goal.

All right. Now to the the last slide available in it. Back to our. Yeah. Thank you so much Rachel. That was awesome. Overview demo. And thanks for answering questions on the fly. So a lot of our resources to answer your questions and to find more questions are on Experience League. So we have some ability for you to scan the QR code. I posted some of these, these links in the chat, and I’ll just follow up and do so again. And then we do have a lot of questions, Rachel, if you don’t mind answering some of these offline, that or sorry. Online. Off mute. That would be great. So I’ve tagged you in a few and I’ll read a few out loud. Whatever’s easier.

Yeah. If you want to, maybe we can go down the the list. It seems like we have some some time left, so we can probably get to all of them. Yeah. That’s great. Let’s see, let’s see. Yeah. So I think I think you kind of answer this question, but just to confirm, can we ask AI assistant to show the audiences that are looking for a specific experience event like survey dot completed? Yes, yes. So that’s where today you can ask with the specific dot notation. So which audiences use. And then fill in the attribute name using dot notation. And then soon will be able to more flexibly not guess but like match your natural language to two different fields. One add on I’ll provide there is we’re also working on a disambiguation experience, so if you I don’t see the exact example, which, what was the field example? It was for. Let’s see, it was for. Oh, yeah. Show audiences that are looking for specific experience event.

And what was. Not repeated. Survey dot complete. Oh yeah I see, you know. So yeah, you could write like, you know what showed me all the audiences that have completed surveys. And we would be able to then figure out okay, which experience event does that correspond with in a, in a case where perhaps you don’t know all of the event names or another user coming in might not know all the event names. So we’ll make it even more flexible. But today that should work. Which audiences use survey dot completed? Yep. I think that’s a good tie in to Karen’s question about, it would be great to have AI assistant answer questions like how many profiles are ingested in a day or week.

Yeah. So how many profiles are ingested a day a week. So more on the observability. So that is a use case that we are looking towards today. Our knowledge graph is limited to data every 24 hours. And so it’s 24 hours refreshed. And it doesn’t it’s not able to pick like within the last week or in the last month for those type of observability use cases, we are actively looking at them and we should have an alpha early next year for that.

Great. And then there’s a question about risk rating. So where can I find risk rating from vendors for you as you Australia regions. And then there’s examples below like general purpose AI, high risk, low risk, high.

I am not sure on that one. Ariel, do you know. I’m not familiar with that one either. So I’ll have to investigate, and see where to find that.

Okay. Yeah, yeah. And then, Mary was ask or actually Joe is asking will assistant eventually add create objects? And I think you did touch on that in the last. Yeah. Yeah, absolutely. And would love to hear more about the types of create use cases you would look for from AI assistant. Again we’ve heard some of the ones around there could be really tedious repetitive create tasks. But but.

Yeah, this one is more about schemas and audiences. Sorry. With that. And then Mary was asking, is there an option to pick and choose one or more of the six features? I think we might need more clarity. Mary, on, what features we’re talking about. And then it’s the will that customer support tickets, will that customer support tickets are initiated through. I get automatically aligned with your instance.

So we’ll cut a customer support tickets get automatically aligned with your instance. Yes, absolutely. So I think if that means, like, are they associated with the sandbox that you’re in? We would have a kind of a Microsoft dynamics integration that would enable you from AI assistant to create the support ticket for case. Yeah. And then, Marianne was asking, I was able to give me a list of audiences use an 51ºÚÁϲ»´òìÈ native attribute, but it doesn’t find attributes that come from our own CDP. Will that resource be added down the road? Oh, it’s there CDP. I don’t think it’s real time. That’s how I’m writing that. Got it. And so yeah, not today. So it’s just looking at audiences within real time. CDP this the 51ºÚÁϲ»´òìÈ Real-Time CDP? Yeah, it’s just with as we described, it’s within the existing, products we have. So it’s within real time CDP Journey Optimizer and within Customer Journey Analytics. So it can only do that for those products. One quick, comment. So going back up to Mary Daniel’s question, I was just rereading it. So is there an option to pick and choose one or more of the six features I think. Is this referencing to when I showed that kind of roadmap slide of these are the six things that are coming.

One answer to that is that you will always granularly granularly be able to enable those features to your, practitioners. So if you’re an administrator and, so there will be that granularity. Oh, here we go. Here are the six features. Information retrieval generate content data analysis and interpretation. Customer support, optimize automate workflows. Yeah, yeah. Cool. And then another question is will there be a guide available coming out to build complex prompts for AI assistant? That’s interesting. Yeah. So that’s a great question. God complex prompts. So today what we have is on the discoverability panel. And then we have an experience league. Guidance on creating prompts. We are working to build kind of a use case library that will help to guide you even further in the different use cases that AI system can help with. And another in product functionality that we’re working on building right now is this prompt autocomplete and related suggestions. So what that will do is kind of train you as a user in the way that AI assistant likes to get its questions. So as you’re typing, there will be a dropdown of five, completed prompts that you can choose from. You’ll be able to edit those as you wish, but that will kind of serve as an active guide into prompt writing for AI assistant.

Cool. There was one question I think I missed earlier. It’s do AI assistant product and operational prompts work on every type of object? Every type of object available on the left hand menu. So for example like offers sources, policy permissions. Yeah. Good question. So for now ask is and just is across those eight different objects audiences attributes, data flows, data sets, destinations, journeys, schemas and sources. We are expanding actively to the other objects. Particularly in the journey space, because today we just support journeys but not campaigns, events, etc… So I would recommend to head over to the discoverability panel inside of the product, that light bulb icon at the top. For the full list of what is supported today and stay tuned for what is coming soon and we’d love to hear from you. What you would like to ask about.

Great. Cool. I’m actually going to post in the chat some of the the AI user guidelines, so that will be accessible to all. I think we did a pretty good job covering the questions. Julia says, are we going to have more series like this? It’s very insightful. Thank you. Yeah. So I think we covered most things, I think going I’ll pass it over to our host and we’ll take it from there.

Perfect. Thank you guys so much. That was great. Very informative. And thank you all for joining us today for tech session. Just a reminder, I will be sending out an email 24 hours from now that includes a recording for this and a link to the Experience League page where you can rewatch this video. Thank you again and we hope to see you guys on our next session of tech session. I.