Jeremy Davis
Sitecore, C# and web development
I've had a few days to decompress and think about all the stuff I heard from the various stages at this year's Sitecore Symposium in Orlando. So here's a rundown of the key messages for me, from my viewpoint as a developer and architect in the Sitecore space.
url copied!
The top news for the week was Eric Stine's big announcements about the new branding and licensing for XM Cloud and the other composable products which sit around it. They're being rolled together into the new unified "Sitecore AI" brand in a "buy one, get to see all the rest integrated" model.
It's a bit of a trope that in IT pretty much every architectural approach that we used in the past will become important again in the future. And the wheel of time has spun around in Sitecore world again - we've gone from XP's "all in one DXP" through XM Cloud's "composable CMS" to this launch of Sitecore AI that they described as going "from composable to composed".
The key thing that has happened here (other than the change of name) is that buying one of their core products (CMS, CDP, Personalisation, Search and DAM) will give you access to limited editions of the other products too. The ones you don't buy will be scaled at "good enough to play with, but not fit for production" sizes which is an interesting move. In the past it took effort to get access to products you didn't own, so if you wanted to learn if they were fit for your work it could be a bit of a hassle. So this seems like a positive change. Less hassle having to speak to your partner / account manager to get access to other tools you want to try.
They also made a big point about licensing being simplified by there being "one metric" that determines the price for a product's license. However I think this needs clarifying slightly from their on-stage message - each of the products (CMS, CDP etc) will have a metric, but they're not all the same metric. For example: CMS is focused on requests, but CDP is looking at the number of contacts you're dealing with. So not quite as simple as implied, but hopefully works out easier for people buying into the platform.
Plus this new branding is wrapped up in a lot of work about better integrating all the parts of the platform. Eventually bringing everything together (both in terms of the underlying technology platform, and the UI) so that you manage it all in one place, and the useful connectivity between systems is built in. This isn't news (I'm pretty sure I've reported "we'll be merging all the admin UI together" and "we'll better integrate the products" from previous conferences) but the model here seems more future looking and technically better for our AI age. Plus the talk is of changes coming fairly soon, rather than the previous announcements where this was further down the roadmap.
url copied!
And part of that "merge it all together" plan is to enable a better approach to AI integration. "Not just AI on top, but AI built in" was the message. And unsurprisingly that topic got a lot of coverage from the main stage.
They've retired the "Stream" brand for the SaaS products now that the whole package has AI in its name. But the features remain and are being extended with a new focus on Agentic AI and the sort of automations that can bring. And there's licensing change here too - they're saying "there will be no billing for AI by 'tokens used'" for their revised offering. What exactly that does mean for billing isn't clear yet, but it seems like it might help drive adoption.
Plus the MCP Server for all of this is available now, so you can integrate your external AIs to Sitecore data too.
But the key new features here was the introduction of their "Agentic Studio" tooling, and their future vision for AI:
The refreshed SaaS package is shipping with about 20 agents that can do tasks around content creation, branding, localisation, marketing optimisation and governance. They intend to extend the features here over time, and provide the ability for you to integrate your own agents from other systems as well. They want all of this to be "human in the loop" to avoid any unintended side effects of LLMs, so they've built UI for collaborative work between people and AI agents.
Plus there are now numerous places in the UI where you have chatbot-style tooling to interact with the system. So if "vibe-content-management" is your thing, then this is your moment...
url copied!
Up until now one of the challenges of the move from the on-prem DXP to SaaS is that cloud software is difficult to customise compared to the classic DXP we've been used to. The move to XM Cloud took away most of the mechanisms for us to modify the editing experience or integrations that we could do before. But the Sitecore AI is bringing back ways to enable these customisations. Dave Tilbury talked about these in one of his sessions:
Now, we have four places where we can apply customisation to our SaaS software:
With App Studio and Marketplace bringing back something like the old "Marketplace Modules" approach that XP had, this should make it much easier to add integrations for editors to access 3rd party data, or automate tasks.
url copied!
After SUGCON EU this year I reported on the session where Sitecore had talked about their work to replace the old container-based data storage under XM Cloud. Well this work has continued, and it seems its scope has expanded a bit.
There's some new search work coming as part of this which integrates search with this underlying data layer. I didn't get to see any sessions which talked about the detail of this, but with the move to a "publishing is instant" data model for content, it makes a lot of sense to integrate search behaviour into this. You want your indexes to be immediately up to date at the point some new content becomes public. And this all goes hand-in-hand with the idea that more of the UI is moving into the current SaaS portal too - search config will follow that path.
Relatedly, it looks like DAM will be adopting this data model too - which should lead to a much cleaner integration for that to take over the role of the media library in the future. And there's also work on a more integrated data model for the CDP / Personalisation side of the product too.
So it looks like we're getting ever closer to the point where the old Content Editor and container-based infrastructure that launched XM Cloud will be going away. And the result is a more centralised data layer which is better suited to supporting the AI services with fast and accurate context data.
url copied!
Sitecore have reiterated their commitment to keeping the old DXP alive as long as their is demand for it. But it's future appears more interesting that I would have predicted in the past:
I'd been fairly convinced that XP and XM would be stuck on .Net Framework for the rest of their days. I was aware of some work Sitecore had done internally many years ago to see what "rebuild in .Net Core" (as it was named then) would look like, and the answers from that didn't seem positive enough that the investment seemed worthwhile. But things have clearly moved on, and one of the sessions this year discussed their plans for the "Post 10.5 world" where they want to start the process of moving towards using the modern .Net to run the platform DXP:
This isn't a simple change, and it won't happen overnight. But they do now have a public plan to transition the software over a series of releases. They're breaking it down into to main bits of work:
The rendering bits
As
Vignesh
said in his talk, the majority of the custom code in an XM/XP deployment is in the MVC UI for your site. Controllers, models and Razor files form the bulk of what you build, and many organisations aren't going to want to be forced to re-write this stuff to keep using their DXP installs. So for people who can't change, they're going to produce a modified .Net Framework-based CD server install which is as lightweight as possible. It will strip out anything CM-related from this install to reduce the size and complexity of hosting the .Net Framework CD. That should lower costs and complexity for these roles.
And for people who can migrate, there will be a modern lightweight Editing Host role to allow you to get rid of the CDs entirely if you're using headless. (Which already has support for ASP.Net Core as well as Next.js)
The back-office bits
For the authoring data processing bits their plan is to use what they call
the "strangler fig" pattern
to wrap the old code with new APIs, and slowly replace/remove the old code so that you end up with a modernised deployment. They're going to try and make as much of our traditional editor customisation into configuration as possible, but for the places we still need to customise there will be a new SDK. This will support a mix of in-process and out-of-process integrations of new .Net code for our custom bits. These will run in a new role in the deployment which uses modern .Net, and slowly this will entirely replace the old CM and related roles.
It's clear that during this change process there's going to be a bit more complexity in our deployments. There will be time where you need some of the old roles as well as some of the new ones. But in theory the result will be a version of DXP which you can run cross-platform natively. And if that comes to pass, running on Linux could significantly reduce hosting costs, as PaaS and container deployments to non-Windows hosts tend to be smaller and less resource intensive.
This does mean DXP will start having to follow Microsoft's support lifecycle for .Net though - so they are going to replace the old Sitecore support lifecycle and align their version timings with supporting LTS versions of .Net in future:
Basically that means we'll have 2 years support for each main release of the future DXP for maintenance and features, and then two years of maintenance (probably security fixes) only. And then the version will go out of support entirely That's shorter than we have today, but one of the project's goals is also to make the upgrade process more streamlined. So they're arguing this will be more benefit than hassle.
This seems like a pretty challenging bit of work, but it could be very useful to the small percentage of the current Sitecore marketplace who can't move towards SaaS products for business reasons.
url copied!
But those were the announcements that most interested me most.
These are interesting times to be working in the Sitecore space...
↑ Back to top
