Sitecore has announced over 250 updates across XM Cloud, Experience Platform (XP), and Sitecore Stream. The biggest part of the announcement is about the introduction of AI capabilities across its products as the company becomes an AI-first business.
The updates to Sitecore Stream come four months after the product was unveiled at the Sitecore Symposium in Nashville. Stream is all about the use of AI and how it can make things easier for marketers and campaign creators.
Updates to Sitecore Stream in this announcement include:
- Brand-aware AI keeps AI-driven actions on-brand and makes them smarter over time, enabling marketers to create more personalized, conversational experiences at scale.
- AI copilots and agents free up marketers to strategize, execute, and scale. Brand, campaign, content, experience, and optimization copilots are now available.
- Agentic workflows harmonize human input and AI automation, balancing automation with authenticity. Levels of agentic automation now include chain of thought prompting, human in the loop, and fully agentic.
Roger Connolly, Chief Product Officer at Sitecore said, “As the marketer’s landscape rapidly evolves, Sitecore is ready to enable them to harness the power of AI for enhanced creativity and productivity.
“Our AI-driven solutions offer customers the ability to elevate their content management experience, and the flexibility to integrate their own AI technologies within our adaptable framework. With our latest innovations across all core products, marketers can now create dynamic experiences rapidly and seamlessly scale their digital presence, delivering exceptional customer value.”
Enterprise Times also spoke with Sitecore CEO, Dave O’Flanagan about the changes, and especially the impact of AI.
Key updates to the Sitecore product suite
Sitecore has broken down the updates into five main areas. All have new AI-enabled components as the company looks to make it easier for teams to create and manage content. A keystone of these announcements is that Sitecore Stream AI capability is available across the whole of its Digital Experience Platform (DXP).
That move will make a significant difference in how the platform will be used. It includes greater tagging of all assets. That tagging allows the AI to search for and assemble campaigns. For large customers with vast asset libraries, this will enable them to get much more value from their assets.
It will also ensure that branding is front and foremost of all campaigns. The new brand management tools provide much more granularity in how bands are managed and defined. For marketing teams, it should remove the risk of creating campaigns that are off-brand or risk brand damage. One key element in preventing brand mistakes is the use of tone while voice-checking.
Another boost for marketing teams is far greater analytics around campaign performance. This will enable them to fine-tune websites and better understand user behaviour, which is key to improving customer experience (CX) with the brand. To further boost CX, AI will help teams tailor the messaging for all customers, not just the most lucrative groups.
One feature that will interest many marketing teams is Sitecore Personalise. It will provide a list of actions for smarter personalisation decisions. As the company moves further down the agentic AI route, it is likely that this will become a fully automated response.
To boost fine-tuning, the new AI tools will allow teams to create, duplicate, and manage websites without needing IT support. This allows brand teams to respond quickly to new opportunities and refocus when circumstances require it.
There is much more detail about all the new updates in the Sitecore Changelog.
What did the new CEO have to say about this release?
Since O’Flanagan took over the CEO role, Sitecore has seen a refocus on its products and some significant changes in the management team. Both should come as no surprise. He is a product man, and that shows in the updates and changes to the products. But to make that happen, he also needed a management team who could make that happen.
He commented, “I wanted to really refocus what Sitecore was known for, so we’ve doubled down. We want to be the undisputed leader and the vendor to power the content life cycle. From a product perspective, that’s doubling down on Content Hub to think about the content operations aspect of delivering content to the customer, and XM cloud or XM XP to be the experience platform to deliver the end user, personalized experience of the consumer.”
To make all of that happen, the company has invested heavily in AI. O’Flanagan commented that the company has made space for an aggressive AI investment strategy. Over the last 12 months, it has been adding AI to its products and focusing on improving capabilities for users. That’s important. It’s too easy to add AI just to keep up with competitors.
Agentic AI redefining campaign creation
Like many companies, Sitecore is talking about the benefits of agentic AI. Unlike the usual AI to do mundane jobs, agentic AI is about asking it to do complex tasks that it will then break down into smaller jobs and then execute. Importantly, it also pulls from multiple sources, such as databases, to complete those tasks.
From a marketing perspective, that might lead to a prompt of something along the lines of – Build me a website and marketing campaign around this brand. Focus it on 16-24-year-olds in the US and Europe. Create subsites that use local language and include images, video, audio and text. As part of that campaign, build content for social media channels.
It’s the sort of opening statement you might expect in a meeting to define a new campaign. But for agentic AI, there are multiple challenges. Does Sitecore’s AI understand multi-model content? Does it understand brand limits, including contracts? Can it structure the content properly? More importantly, where are the human checks and balances?
Addressing the multi-modal challenge
O’Flanagan commented, “Content Hub is, is a critical strategic asset in this conversation. We store textual content, images, 3d video, audio, everything in that solution. A lot of the AI focus initially was on that area due to use cases in that space.
“Image tagging is a good example where we add AI logic to images. It can do machine vision and can then work out what the image consists of. But it doesn’t just apply random tags. It allows you to tag based on the taxonomy that’s defined in the organization. This is really important because it allows you to create structured content at scale.
“That abstraction can equally be applied to video. It can equally be applied to 2d 3d once we define the agents that have the capability to interpret the multi-modal content.
“With agentic, if I wanted to do grounded image tagging and localizes and then be able to publish that content in a specific media campaign, I would set out a workflow that would take in an image and immediately tag it with my taxonomy. It would take that then automatically to a set of agents that could do one or more localizations.
“There may be a follow-on content audit. At some point, there’s a human in the loop to say, Yes, I’m okay that this image looks like this, and it could flow down the chain to ultimately be activated in a campaign”.
Interestingly, O’Flanagan then said that the Sitecore wasn’t quite ready to put that into full production yet. That’s important because it gives a view of where the company is going. It also shows that some of the updates are foundational elements for this future.
The Human-In-The-Loop
O’Flanagan was quick to point out that Sitecore already has human-in-the-loop (HITL) technology in place. It starts with how the brief is defined because that has to come from campaign teams. What will be interesting is to see how complex they make their prompts for the AI. The more detailed the prompt, the better the result will be.
For now, the agentic AI will collect all the content that it believes is related to the campaign. It can push content to certain channels, including social media. It will also build the websites and subsites without the need for IT. That update is part of the 250 in this release.
However, that is where humans step in. At present, the AI cannot push directly into the websites without the content being checked and verified. Interestingly, despite the on-IT website update, O’Flanagan said, “The step from there to the website isn’t automated, but will be soon.
“I would assign a number of tasks to the website operations person. They would go in and start using AI in our CMS to create the create the landing page, the imagery or the content. They would take the images and content we sourced from the campaign. It’s already there for them and finalize and publish.”
For O’Flanagan, the agentic automation is currently about 70% of where it wants it to be and they are already able to demonstrate this. By Orlando in November at the next Symposium, it might well be that we will see this at 100%.
The need for contract management
Content is a tricky thing and is subject to copyright laws and contracts. Many organisations license it for specific campaigns, and there may be usage limits, including what territories it can be shown in. If artists or models are involved, the content may be limited to one specific campaign and for a specified period of time.
All of Sitecore’s customers will have large legal departments that will want an overview of campaigns today. They will want to verify the rights to prevent damages and impact on the brand if they have to withdraw a campaign. So how will the AI understand rights if there is no contract database linked to the CMS?
O’Flanagan made it clear that customers are going to build their LLMs on the content that they have. However, he also said that customers haven’t yet shared concerns over the risk of the agentic AI creating a contract breach. He believes that this is where the HITL will act.
That is surprising. One of the fastest-growing areas for AI has been the contract market. It was quick to adopt machine learning a decade ago and AI was a natural step for it. It raises a question as to how long it will be before Sitecore either acquires an AI contract vendor or builds its own. The obvious solution would be acquisition, and it will be interesting to see how O’Flanagan reacts once customers ask about this.
Enterprise Times: What does this mean?
The last 12 months have seen a very different Sitecore emerge from the one that O’Flanagan took over. He wants the company to be known as an AI-first business. From this latest announcement and his comments as to the focus of the company, that is well on track.
It will be interesting to see how quickly customers adopt all the new capabilities. For many, especially those with very large content assets, the tagging process will take time and require investment. However, the benefits that the company promises from that effort, are likely to far outweigh the costs.
For O’Flanagan, the increase in AI should not be seen as a threat but a help. He says, “There will be a significant role for marketers QA in the process so that people can use it confidently.”
