Document Authoring at Scale: The Foundation for Intelligent, AI-Optimized Experiences
Discover how Adobe Experience Manager enables scalable, AI-ready document authoring across global enterprise ecosystems. Learn how integrating AEM Assets, Workfront, and translation workflows creates content variations for markets and languages, forming a foundation for semantic enrichment and AI optimization. This session features insights from Andreas Haller, Principal Solution Architect at Cognizant Netcentric, recorded live from Basel. Explore how Adobe’s ecosystem supports intelligent experiences, ensuring alignment with corporate governance and enhancing digital transformation strategies.
Let me be here. This is about edge delivery services, specifically in combination with document authoring on DA Live.
To wake you a bit up, first of all, who of you has either said or heard this before? So raise your hand. On top of that, who has not only heard it but believes it’s true? I think you’re a bit timid today, but I guess we have a few.
This talk, I’m not saying it’s true or false. I want to tell you my experience with it. Because what we have done, we have one of our dear clients, 230 regions, nine languages over them, many integrations like the stuff like login, external search. You heard this in the internal search now, but this is still external integration. We have prices from commerce, etc. And we have tens of thousands of products. So your typical EDS project. So I didn’t bring you a demo today, but I at least brought you a few screenshots to look at it. So we decided for DA, let’s start at creating our content. How does it look like? You might ask yourself why DA and not Universal Editor or something else. This is one example why I show you a couple of more, but we wanted to use Smart Drops, Dynamic Media. And DA has actually enabled us to, after we select our assets, the nice assets, like you probably know, to select your Smart Drops right in DA. This is an extension we provided. It’s out of the box now in DA. And this kind of open source, we are able to extend it and extend it actually with the same technology we have on the actual front and on edge delivery. That’s one of the key factors we decided for DA. This is the way forward, but it doesn’t stop there. We heard it a lot from Gabriel, but also from San Jose. We need to be AI ready. DA, quite a good basis for that with edge delivery services. I’m not even talking about ease of use, of document authoring. I mean, there are many opinions out there, but it’s still a good option in my opinion.
Rapidly growing feature set. I mean, you saw on the slide before one we contributed with our project, but there are also many others starting now, and we believe that’s a good basis to build on.
Besides the good, always there are the challenges. I want to share the experience.
Change management, especially with the authors.
They are used to maybe the AEM, what you see is what you get, interface to click, maybe even a completely different CMS structuring the data. DA is quite a cultural thing to get out there to them. They need to unlearn a couple of things. They need to be open to the change. That’s one of my biggest things I want to tell you. You need to talk to them, make them understand what’s the advantage, disadvantage, and also understand them where they are coming from so that they can understand what you’re doing and what you’re providing to them. Metadata editing, probably one of the biggest disadvantages of document authoring in my opinion, metadata is just the form base, and in documents you don’t have that. DA has already a couple of nice starting points to make that better, but I think this is one of the things I’m looking most forward to in the next month and years to make that really a great integration in document authoring. Bulk editing and reporting. Also DA has a couple of very nice things, especially reporting. We did already a couple of POCs. I mean, we have the lab afterwards, like in the afternoon, so you see how to extend DA reporting is also another great thing to do extensions in DA because they are most of the time quite specific.
So this is the DA side. What about the general EDS development side? We merged from a feature branch to production, said no enterprise architect ever. So sure we are bringing out this, okay, we are super dynamic, but now we have a company that tells us, listen, here our standard is we have a product environment, a staging environment, a integration environment, a whatever, like the list goes on, you know the drill.
So how do we do that? Our approach is start small, start with the EDS kind of minimum in a way, right, like the typical EDS where you would do, but then ask them, okay, what is it you really need and what is it you’re just doing like this because it was always like that, right? And you will end up with some compromise. It won’t be feature branch to production, but it might be something like that, right? We see on the right hand side here our development set up in a way, right? We have like what you typically know, we have feature branches, we have a particular main branch, and we can directly test on those feature branches, right? But we also have a lot of testing going on here on the main branch and not on the feature branches because we have all those integrations. We have the login, OIDC, OAuth needs those fixed redirect URLs, right? So this is everything non-prod, right, on the right side, but it’s still highly dynamic, nice to develop. You have to have still your nice developer experience on your local, but then there’s also the production, right, where we have like UAT happening here maybe even, right, and then at some once a month maybe even we did more rarely, right? We are moving over to the production environment. So I think this shows quite nicely you can have a lot of advantages, maybe not all of them, but it fits into an enterprise environment. You adjust and add what you really need, but maybe not lose all of the flexibility, right? Like coming from there and then like slowly add what you really need to make the processes, et cetera, maybe even the legal requirements, right, work. So having talked about environment setup, back to the content.
So multi-site management, right? I said 230 regions, nine languages. How do we duplicate our content, right? This is how MSM works. It copies everything.
You don’t, right? This is nothing we in this project or I invented.
This is something with actually if you look a bit into the am.live documentation, you’ll see some concepts about this, how to do this in Edge delivery service, but I wanted to mention it because it was the first time at least I have really applied this in scale in such a big context.
And how it works is you have the user accessing your website. Your user typically has some preferences. Either they are logged in, right, or they’re coming for specific IPs. So you probably know what this user’s region, but also language is. Typically websites like have a default. I’m coming here from Switzerland. So then my default might be Swiss German or Swiss French or whatever.
And now, like with every user request, what we implemented is we are finding the best matching content. How we do that is all our URLs start with a locale. Locale can be just the language like D-I-T-F-R-E-N, right? But could be also country or region specific like EN-CH, right? And then the rest of the URL, this is what we are calling our locale agnostic portion of the URL. And with this locale agnostic portion, we are actually going to our query index in ES and looking now for a best match. So the best match if I’m in Switzerland and I’m speaking German is obviously the German-Swiss content, right? But if that’s not available, we can define a fallback path. Maybe next I want to look for global German content, right? Or then even if that’s not available for global English content, and only if those are not available, then I’m showing the 404 obviously, right? But chances are typically I’m falling back to the best match. And I’m also falling back in the URL, right? So that the content is not duplicated.
I’m really showing them if this piece of content is only available for global English, I’m going to use it. And obviously this has some challenges if you’re thinking about multi-domain, but at least if you’re on one domain, you can have quite nicely a duplication of content without this copying. And also DA supports quite nicely this approach. As I said, I didn’t invent it. It’s on AM. But I documented how to do it, also how to rewrite the links on the page, right? Because you might want to show the global English content to this user, but all the links in this context should still point to the next best version of whatever other content we have there, right? Looking more into the backend supply chain, you heard a lot about the AI also there. The A has for now proven, we are still in the making of integrating a couple of things, But Workfront, right? Like all the asset generation, also content generation, we are orchestrating in Workfront and SDA also API first and provides quite good interfaces for that. We have for now quite a good process for that. Assets I said already, we are integrating with dynamic media. Also the smart prop, etc. And translation there is quite a nice out of the box UI in DA for handling that. Next topic is commerce.
Commerce is prices, etc. This is kind of integration API calls, right? But one topic which is always a challenge, there is a couple of things coming up, like the commerce orchestrator from Adobe, but for the pages, right? Has also changed over the last couple of months and years in EDS, the approach there. What we are doing is quite a custom approach because we have many, many custom content So we have not only DA and assets, we have also a pin with the structured data and we have commerce with a couple of eligibility rules we need to apply.
And our process is starting at the top, we are looking at DA, we have one product we want to render.
We are getting a template from there, we are getting all the marketing content also from DA and basically this happens on preview click. We are assembling a whole product HTML and pushing it to edge delivery and from there on every product detail page is like every other edge delivery page with nice performance scores etc. This is our commerce. As I said, we are in the middle of the project, so we are close to getting live with the first results, but there is already now and that’s because we have DA edge delivery and there’s even more ideas to come, right? So we feel we have really a head start, so we are very flexible, we can introduce all those new AI capabilities. Obviously, we need to do a lot of unlearning, question the status quo. This is also a fight sometimes, right? And you need also the customer to be open for change, but it really pays at the end. For us, developers, architects, the challenge is really thinking outside of the box too. It’s not only the author saying, okay, where do I click to get the page properties? It’s also us saying, okay, where is my depression environment and where is this and how does it work? So we also need to step one step back and have a look at the greater picture, what’s our actual goal here and how can we achieve it. What we are planning to do is more extensions to the DA live, the product pages, and we have only POC at the moment, but we will also bring this live during the beginning of next year. Keeping performance high, sure, you have this 100 Lighthouse score, right? But we all also know it’s a constant investment. It might not be as big as with other platforms, but still we need to keep it. We don’t need to forget about it, right? Because we heard now about all these great things in DA, product integrations, right? So the core goals of edge delivery service, it’s important to not lose track of them. We need to still have them in our development cycle and in our goals. And yeah, this is a very important point like the last one. Now we have really the basis to invest in the supply chain and in this ease of use AI for the end users so that our pages are actually found by the AI, but also for the authors to generate the content. And with that, thank you.
Thank you.
This session — Document Authoring at Scale: The Foundation for Intelligent, AI-Optimized Experiences — features Andreas Haller, Principal Solution Architect at Cognizant Netcentric, recorded live from Basel. Discover how Adobe Experience Manager Document Authoring supports global, AI-ready experiences across enterprise ecosystems. Learn how scalable, document-based foundations integrate with AEM Assets, Workfront-powered supply chain processes, and translation workflows to create content variations across markets and languages — forming the base for semantic enrichment and AI-driven optimization.
Special thanks to our sponsors Algolia and Ensemble for supporting Adobe Developers Live 2025.
Next Steps
- Continue the conversation on Experience League
- Discover upcoming events