Publishers have long been told their content and first-party data are their most valuable assets. And, as the agentic AI craze takes hold, publishers have been inundated with advice on how to adapt these assets for AI solutions.
But what does it mean from a practical standpoint for publishers to prepare their content and data for the AI era? For TIME, it meant a long overdue overhaul of its content management system (CMS).
On Thursday, TIME announced a switch from WordPress to Contentstack, an enterprise CMS. TIME spent four months this year migrating its more than 100 years’ worth of content to the new CMS to make it more easily digestible by TIME’s consumer-facing AI tools, as well as third-party solutions, said Michael Mraz, general manager of B2C and chief product officer at TIME.
The CMS migration helped unify TIME’s fragmented content data after years of platform transitions under multiple corporate owners, Mraz said. This unified data foundation enabled TIME to launch its own AI search product, create new interactive experiences on its site and convert its archival content into “markdown” pages that are easily parsed by large language models.
Content infrastructure
Like many publishers, TIME is experimenting with ways to use AI to keep readers engaged on its site. It’s also trying to make its content more visible in AI search interfaces and more frequently cited by AI agents.
Mraz joined TIME last year to head up development of its AI platform. One of his first projects involved working with AI software company Scale AI to create the TIME AI agent, a chatbot on the TIME website that pulls its answers from the publication’s vast backlog of reporting.
But, in designing this tool, TIME realized that years and years of slapdash CMS platform updates and inconsistent content classification had left it with a fragmented database, Mraz said.
In the past few decades, TIME had undergone multiple changes in ownership. The company was acquired by AOL Time Warner in 2000, then by Meredith in 2017 and finally by Salesforce owner Marc Benioff in 2018. Over that time, TIME had also cycled through “many different tech and product teams,” Mraz said, “and the resulting CMS situation was incredibly complex.” Plus, TIME’s last major CMS update was in 2014.
As a result, “we had four different content templates across four different frontends, competing code bases, unstructured data all over the place,” Mraz said. Which was not exactly a great foundation to build AI products on, he added. So, to ensure the TIME AI chatbot was being fed with consistent data, something needed to change.
“We were still on WordPress at that time, and the process of working on the agent was a forcing function,” Mraz said. “In order to be able to develop and iterate and innovate, we had to be in a different platform.”
Also within the past year, TIME had kicked off an initiative to create markdown versions of all of its content in partnership with publisher AI licensing marketplace TollBit; these markdown pages only include metadata that is easily understood by AI software, in contrast to rich content designed for human readers. Mraz realized this process would be “vastly simplified” by cleaning up the CMS backend.
So Mraz began looking for a new CMS partner that could handle migrating TIME’s massive archive over to a new system. Ideally, he added, it needed to be a CMS that was built to prioritize AI and API integrations, rather than a legacy CMS that wasn’t designed with AI in mind.
Prioritizing API integrations was particularly important, Mraz said, because, previously, TIME’s development team “was playing whack-a-bug” and trying to address coding issues as they arose during routine manual software updates.
Switching to an API-first platform, he said, made it easier to automatically roll out updates from various content syndication partners and TIME’s email newsletter platform beehiiv. It also streamlined integrations with TollBit and other agentic AI partners.
Mindset shift
TIME decided to go with Contentstack not just because its platform was built with AI and API integrations in mind, Mraz said, but also because he was impressed with the company’s customer service.
To make the CMS transition work across such a large enterprise as TIME, he said, he needed a partner that could guide editorial and management through the process. That partner also had to work well with TIME’s web development agency Code and Theory, which, last year, helped overhaul the publication’s website to make it more interactive.
Assisting publishers through such a transition is as much about promoting a shift in mindset as it is about technology, said Neha Sampat, founder and CEO of Contentstack.
“With all the AI stuff that’s being thrown at everybody, from the reporters all the way up to the people making the technology decisions, change management is real, and it’s stressful,” Sampat said. “So, part of the partnership is not just making sure that the software works and that we’re fixing things when they break, but that we’re actually in conversation with the team as they’re transforming.”
For example, Contentstack had a dedicated product manager work with TIME’s editorial teams to identify what tedious parts of the publishing process could be automated in order to improve “speed-to-publish” times, Mraz said. This included creating in-house workflow optimization agents, such as one that handles content tagging.
These internal workflow improvements are already paying off, Mraz said. He pointed to new analytics tools TIME has developed that are better able to pull in data across its wide array of content and different end user touch points like email and desktop, since everything’s now unified within Contentstack’s CMS.
These analytics tools are helping TIME editors make more informed decisions about what content is resonating with audiences, Mraz said. Which is especially important, he added, as publishers are seeing less traffic overall due to the rise of zero-click AI search chatbots. So, they need better data on how to drive direct traffic to their owned and operated properties.
“My north star in thinking through these big product challenges is that so much comes to discoverability,” Mraz said. “There’s so much content out there, and being able to personalize and serve the right content that they’re actually interested in, in the medium they find most engaging – it doesn’t have to be much more difficult than that.”
