[Music] [Monika Holdak] Question. Don't you feel sometimes, especially at this type of events like Adobe MAX, a bit like Alice in Wonderland or stuck with this whole AI magic. It's a kind of excitement and anxiety at the same time, isn't it? [Pascoe Morrissey] Yeah, absolutely. I think the excitement definitely comes from all of the new doors that are being opened for designers and all of the amazing tools that are now available. But at the same time, I think it's definitely daunting. I think a lot of people perhaps feel like they don't really know where they are. It can be easily feel like you're lost in a world of tools that don't make much sense. But hopefully, today, we can put some of that to bed and teach you some of our learnings about what we're doing to get the best out of AI. So I'm Monika from Deloitte Digital.
And I'm Pascoe Morrissey.
And we would like to invite you for the journey through the AI, just showing you our approach, how to pass successfully this AI looking glass. So let's do that together.
So, Monika, which voice would you be taking in this session? Yeah. As you can see, it would be a session for two voices. So I'm Digital Marketing Strategist, and by heart, I'm also innovation geek. So I would like to show you a perspective of the business, perspective of brand owners, and digital marketing directors, how to deal with those challenges and how to use AI in a good way. And I'll be representing the voice of designers having spent countless hours and lost a lot of my hair resizing images, renaming file types, and things like that. I fully believe that AI can free up a lot of time for creatives to be more creative. First challenge, more business oriented is what to do to increase actually the personalized experience. I mean, how to increase also content production at scale, let's say by 5 or 10 times, at the same time keeping the same cost and capabilities. So we are looking a magic function to keep us with the track, with the pace for the consumer hearts, at the same time, keeping our business in a good shape. And the second challenge is-- And I think from an individual perspective, the challenge is how to make your craft more meaningful for your business and how to get the most out of not only your existing assets, but also new ones, and how to ensure your assets can be easily imposed and decomposed for maximum efficiency and effect. Yeah. So those two challenges surprisingly have the same solution, actually be solved by the same, as I said, magic potion. And surprisingly, the answer is super simple. It might be well described asset.
So why it's so important? Because AI creature needs to be fed by data and doesn't matter if this data comes from IT department or marketing ones. So actually, AI, GenAI engine like Adobe Firefly, treat your beautiful key visuals or your brilliant storytelling as a source of data, and finally, can provide a data as a outcome. So before we get into how to tame this AI beast, just a quick look at what we're hoping to cover in today's session.
So firstly, we will look at the way in which the output from your data is really only as good as the data you put in. We'll also look at the fact that a lot of brands today are dealing with legacy assets that are poorly organized, and we'll explore what we can do about that. And finally, we would like to have a look at how you can get your brand AI-ready for personalized customer experiences going forward. And we will then leave you with three key tips that we feel you can start putting into practice after today's session, so make sure you stick around for that. So, Pascoe, to your first point, could you explain us what metadata is and what it means that we should have a proper described asset? Yeah. Absolutely. So when we're talking about data in relation to all the content, we are, as you say, talking about metadata. And while it might sound boring, it's actually super fundamental to everything we're talking about today. And put simply, metadata is data about other data. And the other data in this case is digital assets. So this includes things such as title, description, date of creation, and who created it. It might also cover things like licensing information, expiry dates, and so on. So yeah, this metadata underpins what we're trying to convey today, which is that in order to get the most out of AI and your assets, you will need to have accurate and consistent metadata in place.
Let us quickly show you an example of this in AEM with an initial look at the accelerator that we have built in order to allow us to populate some of this data using AI.
So here we are in AEM, and we're looking at our assets. And for this purpose, we're just going to click into one of them. And as you'll see here, this is where we start to understand what we're talking about in terms of metadata. So we've got title, we've got description, we've got fields such as file type and language. And also across the top there, you'll see all the different tabs, which again provide really, really detailed information, all of which are related to this asset, and trying to describe it as best it can. In this example, you'll see that there's no title or description currently, but using the tool that we have put together our asset booster, with a simple click of a button, we're able to generate the title and the description in a couple of seconds, save that, and then we go back to our asset view with our newly created metadata.
Thank you, Pascoe. It's very, very interesting. So could you tell us why metadata are so important? Metadata really is more important now than it ever has been before. And I think this is definitely driven by a change in consumer behavior. I think, historically, all users were consuming the same content in the same way. Whereas today, users want to be serve personalized experiences across multiple formats and at a much faster pace. I think this change of behavior means that organizations now are no longer able to just churn out as much content as possible to a blanket audience. Organizations need to produce the right content for the right people and importantly at the right time.
And in order to do this, AI systems are being used to keep pace with this content production. And again, back to that key message that with this increased use of AI, it's only going to work if you're putting the right things into these machines. So there's no use getting your hands on fancy AI tools to churn out images and videos for you if the information you're going to be feeding into these machines is poor. So historically, I think a lack of understanding around metadata and its benefits has resulted in a large number of organizations now facing an asset library with poorly described assets. And I think that's a big problem for today. So it's not only about the right content, but also the right description of the content. And actually, we can talk about the right amount of content, right? Yeah. Absolutely. It's actually very, very important topic from the business perspective because we are not talking only about individual art craft. It's more about a situation when we'd like to control production of the creatives, right? We call it, in Adobe that it's a control of content supply chain. So starting from the planning of the campaign via triaging importance of each of them across different teams and or sometimes different agencies, right? Then we need to understand how to support in terms of GenAI, also content production, how to control distribution, and find finally, how to create a feedback loop to protect and learn and create good analytics and reporting. So actually, the asset management and well-structured asset repository is the key not only at the individual level, but on the organizational as well. And it might be really, really, really beneficial in terms of cost efficiency, in terms of team efficiency, and actually, the campaign, the creatives effectiveness, yeah, and the impact on the end to user journey and user. It's a very good starting point to talk about the issue. Most organizations are struggling with the asset management in general. Let me share the story of one of my clients who was really, really ambitious and put really big effort to do a proper asset transformation while replatforming. So the situation is that everything is in one place. All agencies has own data, own assets, every single type of file open and close one in place, and it was very strong demand to show the outcome using Adobe Firefly and using all this magical toys, magical stuff to increase rapidly creativeness and efficiency of the team.
And it was really, really surprised, negatively surprised, I would say, that he couldn't find this effect what he wanted.
The thing was that the people at the process level and the asset and the inventory level were not well prepared. And it comes to my mind the story with the man who prays and prays to his gods to win million of the national lottery. He prays week by week with no results. Same thing. And finally gods comes and say, "Listen, give me a chance and just buy the ticket." So it's exactly the same situation. You need to do as an organization, as a team, you need to give the features, AI features chance to work giving this proper asset description. Yeah. Absolutely. I think, as we've talked about many times before, I think it's easy to say what needs to be done. It's much harder to put it into practice. And I think we've seen this with a number of our clients, myself, yourself, right across our business. And I think it's the understanding the importance of metadata and well-organized assets is what led us to create the asset booster in the way that we have.
I think earlier, we showed the example of creating metadata for a single asset, which is useful for an individual.
But the reality is that most organizations are actually dealing with vast, vast numbers of assets. And so really, we need-- Like it was in this situation, so actually, the real issue was that after the replatforming and transformation of all asset in one place, the whole place was extremely messy. Legacy assets were not described at all. We found many, many duplication, outdated content, not compliant in every single time. So the first thing what we can do, we need to do a asset data, metadata cleansing, right? And how to do that in smart way, Pascoe? Yeah. So as we looked at earlier for a single asset, we've also developed a method and a tool for doing this in bulk. So I'm going to jump back into AEM. And I'll just give you a walkthrough of how we can start to generate some of this metadata for multiple assets at the same time. Cool. So here we are back in AEM. And as you can see, we've got multiple assets here, which while they do have file names, which are quite non-descript set of words, they don't really have any meaningful data that would enable them to be found more easily through search, for example. So what we're able to do now, with our accelerator is select all of the files in a specific folder, go across to generate metadata. At this point, you've got a choice between title and description or both. In this example, we'll go with both. And with a click of a button, we confirm that any existing data will be overwritten. Hit Continue. And then in a couple of seconds, we have a whole new set of newly named assets.
So here you can see they've all been renamed with names that are far more descriptive, which makes discoverability a lot better, makes searchability a lot better. And all of these things feed into the reuse of assets that the avoidance of duplication.
And really, it's just about making sure that people can find these assets and not just people, also machines. It's definitely works both ways. Absolutely. So it's like the butterfly effect. Have you heard about it? It's like the fluttering of the butterfly wings, tiny, tiny change, which assets description is, and we can get a really, really big impact on your organization. And it's not only about the metadata description, it's something more. Thanks to that, we can learn search engine to make our assets more findable and searchable. Isn't it, Pascoe? Yeah. I think-- I know that you have something extra, to put to-- - Yeah? - Yeah. I think when we talk about finding things and labeling things and categorizing things, the taxonomy is one of the main things that we discussed. And I think it's perhaps not the sexiest part of this job, but it's definitely one of the most important, I think, if you're able to have a...
Strict taxonomy for all of your assets, it makes the labeling process so much easier, it makes categorizing things so much easier. But the problem is yet again, the quantity of assets now and the quantity of tags now means that managing these taxonomies can be very, very time consuming.
However, in the same way that content lacks metadata, it can also be said for content lacking tags and a taxonomy in which to store them. So this was another use case that we found for our asset booster, and I will jump back into the asset booster just to give you a quick idea of what we're talking about.
So here we are back in AEM. We're able to navigate to our Deloitte Asset Booster, and one of the features in here will be Generate Taxonomy. So upon clicking that, we are then taken to this screen, which allows you to specify the folder from which you're hoping to create this taxonomy. And what the tool will do at this point is when you take Generate Taxonomy, it will look at every single tag on every single image or file in that folder, and it will then decide on the most sensible way of categorizing all of those assets-- Sorry, all of those tags, so that you don't have to spend hours and hours and hours manually putting all of the photography tags together, all of the landscape tags together, and so on, and so on, and so on. So we'll go ahead and hit Generate taxonomy, and we'll see what comes back.
Great. So there you have it. Within a few seconds, we've got all of our tags grouped together by different categories. This taxonomy can then be exported and edited manually before reuploading and apply to a specific folder further down the line. Okay. So as Alice in Wonderland says, "It's no use going back to yesterday, because we are in completely different situation right now, and the future with human and artificial intelligence is inevitable." So now we know what type of challenges we are facing. We know how to deal with that. We know how asset description is important. So let's sum up what we need to be sure that our brand is ready for AI journey. So let's sum up what we can do to have right content and well-organized content in our asset library. First point and our prerequisites, mandatory prerequisite, I would say, I would stress, is the single storage location for all our assets, both open and final ones, right? Why it's so important? Because we would like to have consistence across all products, all campaigns, and actually have a one consistent repository for the future, just teaching our Adobe Firefly and other AI engines, our corporate language, our tone of voice, etcetera, etcetera.
Of course, at the end, we hope that we can get efficiency gains and final effectiveness. Cool. Thank you, Monika. So I think when we're looking at how brands can get themselves AI-ready as it were, I think a second point to touch on here is accessibility and legal compliance. I think these two items are both as important now as they have ever been. And I think it's very, very important to get these nailed down early on to avoid a massive repeat of effort later down the line. So if we're talking about training AI models and the data that we're feeding into that model isn't compliant, then funnily enough, all the results that we get will be non-compliant as well. So I think, again, really, really important to stress that accessibility and legal compliance are of vital importance early on in this process. Absolutely. In other golden rule in the digital world is if something is repeatable, actually might be automated, especially when we have some marvelous tools around us. So if we have everything in one place and all assets are well-described, then we can identify some patterns across campaigns, across the marketing or advertising calendar. And if we can find this type of patterns, then we can start reusing the historical data just to save our time and money. Yeah. Absolutely. I think the next thing to touch on is what we've talked about quite a lot today, and that's having a robust process for tagging and metadata creation. I think we've seen, how important this is to everything that we're talking about, and making sure that you have that process in place and everyone's aware of it and it's being followed strictly, not as a side of desk activity. This really needs to be up there with the most important tasks that are happening in the content world, so making sure that tagging and metadata creation takes place. Absolutely. And last but not least, best practice for asset re-use and asset morphing. So everything, right now, what you just said is just to create a good foundation for AI-driven features and automation and boosting actually creativity, thanks to Adobe Firefly. And just to show you how it works in our Deloitte Digital environment, we would like to also present you our Orb Generator. Orb actually is our brand signature. You can see everywhere when you are at Deloitte. So based on our historical data, you could see actually the presentation of the Pascoe. So thanks all of this historical data and assets really well described, we could teach our orb generator, our engine, AI engine, to create orbs just per specific purpose. So right now, I would like to show you how it works for us.
So we can type a very short description what should be embedded in our orb. And in a minute, magic can happen.
And here we go. We have our lovely orb for Adobe MAX. So as you can see, outcome really counts when we are self-disciplined in terms of asset description. Just to wrap up and relook at what we've covered today. So we've looked at the importance of data while feeding these AI machines. We've explored some of the challenges that businesses are facing today with regards to their legacy assets and assets more generally. And we've discussed how you can ensure that your brand is AI-ready going forward.
So we start from the filling description that Adobe MAX, it's a mixture of excitement and anxiety, right? Because it seems at the first side that it's impossible to implement to our organization and to our daily life. But the only way to achieve the impossible is to just believe it's possible, and we, again, very, very self-disciplined. So before we go, we have something for you as a free life hacks and takeaways. Pascoe, what is that? Yeah. So I think the three key things to take away from this, which you can hopefully implement as soon as possible, are firstly, getting all of your digital assets into one place for the reasons we looked at earlier. Secondly, install a robust process for asset tagging and metadata population, and C, start to train Adobe Firefly with your new metadata-completed assets.
Okay. So this is pretty much what we prepared today for you. Thank you so much for being here with us. I hope that you enjoy this session. And if you're feeling like Alice, keep these things in mind for your creative journey. Thank you so much. Thank you. Thank you for watching.
[Music]
