Open the homepage of any digital experience platform vendor, then open that same homepage from 2021 in the Internet Archive. There’s a fair chance you’re looking at what reads like two different companies.
I pulled yearly homepage snapshots for 24 content and experience platform vendors, from 2020 through 2026, wherever the archive had them. Twenty-three returned usable results. One vendor has worn six identities in six years: Jamstack, then composable DXP, then a composition platform, then a visual workspace, then a plain DXP, and now an agentic DXP. Another ran five. Two more ran four and three.
Four of the 23 kept the same category term throughout. One has been saying “The Content Marketing Company” since 2020. Two called themselves a headless CMS every year, though the wording changed. The fourth said “Information Management Solutions” for years and now reads “Secure Information Management for AI.”
The category is louder than it is different
Someone has measured the sameness underneath this. Peep Laja, CEO of the B2B message-testing firm Wynter, ranked 445 category leaders across 61 B2B categoriesby how much of their messaging also appears on competitors’ pages. His finding: 99% of B2B software homepages claim AI, 86% claim to save you time, and the average overlap between category leaders is 57%.
Earlier this year, his team ran a sharper version. They took the value-proposition copy from the five biggest CRMs, removed the company names, and asked 100 B2B marketing leaders to match each to its brand. Seventy-three percent were VPs and CMOs.
They scored 1.86 out of 5. Random guessing scores 1.0. About 43% got one or zero right. Afterward, 74% said they were “not very confident” or “just guessing.”
Laja’s conclusion: “Your positioning isn’t bad because it’s wrong. It’s bad because it’s interchangeable.”
The underlying behavior is much older than the AI wave. Scott Brinker, who publishes the marketing technology landscape graphic that maps this entire category, described it on the Customer Data Perspectives podcast in June 2022.
Talking about how hard it is to sort software into categories, he said categories are “a way for an analyst to think about an industry” and “companies don’t care about those boxes.” When the host said vendors respond by inventing a better category name, Brinker replied: “Everybody wants to be their own category creator and have a category of one. It’s a mess.”
He said that four years before “agentic” showed up on these vendors’ homepages.
That settles how these companies sound. It leaves open a different question, and it’s the one I wanted answered: Does the name still describe the product?
10X your SEO with Semrush for Enterprise.
The world’s most powerful SEO platform, purpose-built for Enterprise.
Request demo
I checked all 17 twice
I took every vendor in the most recent edition of Gartner’s Magic Quadrant for Digital Experience Platforms, published Jan. 28, 2025, and checked each vendor in two ways. First, I looked at what the homepage says now. Second, I looked at what its published release notes say it shipped.
Here’s where they landed.
- Eleven of 17 lead with AI or agentic language on the homepage. Three are running near-identical headlines: one promises an agentic platform for humans and AI agents, another an agentic platform for growth and governance, and a third says it’s built for humans and ready for agents.
- Thirteen of 17 are shipping AI capability that I could verify. Three are not, and for one, I couldn’t find evidence either way.
- The homepage label matched the product in 11 of 17. It overstated in three and understated in three.
The shift is real, and that’s what makes the naming behavior so strange. When nearly every company in a category is building the same capability, putting it in your name tells a buyer nothing.
The overstating cases are where the label gets ahead of the product.
- One company calls itself an agentic experience platform. In the 53 changelog entries it published in the six months after adopting that name, only four mention AI at all, and the single agentic release is a minor update to a component first listed in late February, the same week its homepage changed.
- A second calls itself an agentic DXP and has zero mentions of AI, agents, or the connection standard that lets outside AI tools work with a product across 13 core release notes covering a full year. Its agentic modules exist, separately versioned, and marked beta.
- A third leads with “AI-Tailored” and has none across the platform release notes I could read.
The understating cases matter as much, and they get overlooked.
A market leader whose product page title says nothing about AI shipped a release stuffed with it: content authenticity labeling, agentic permission management, AI search, AI-generated video captions, and an agent’s beta program.Another, whose homepage reads, “Marketers deserve tools that work as hard as they do,” has 11 changelog sections devoted to its AI assistant and the connectors that let outside AI tools work with it.
Three more make no AI claim and show little or no AI work. One has zero AI mentions across 27 releases from mid-2024 to mid-2025, another has none in its recent public release notes, and the third’s only AI item this year is a connector to someone else’s AI tool. They’re the consistent cases, and they’re why I trust the rest of the table.
The excuse doesn’t hold
Ask a vendor why it renamed itself, and you’ll get a version of the same answer. We had to. That’s the category now. That’s how we get into the analyst reports.
In May, Gartner published “Innovation Insight: Agentic CMS” and named six representative providers. One of them has called itself a headless CMS on its homepage every year since 2020, yet has never adopted the new label. It got named anyway. It also added AI agents to its product in September. A second vendor that kept the headless CMS label, and wasn’t in the report, ships the same kind of connector.
The rename was never the price of admission.
There’s a real cost on the other side, and the vendors who held their names are paying it. Analysts organize research by category. Buyers search by category. When you refuse to adopt the term everyone else is using, you’re harder to find exactly when a shortlist gets built. Holding your name is a bet that being findable under your own description beats being findable under someone else’s. Not every company can afford that bet.
Gartner published its agentic CMS report on May 12. One vendor sent out a press release on May 21. Its homepage headline changed between May 26 and June 1.
Nine days from analyst document to press release. Under three weeks to a new homepage headline. Those nine days changed the vendor’s label. Its customers still had the same problems they had on May 11.
What to do with this
Category labels used to carry information. A vendor called itself a CMS, and you learned something about the product. That relationship is weak enough that you should stop treating the label as a filter.
Three checks, all free, all doable before your next vendor call.
Put the vendor’s homepage into the Internet Archive and read the page title once a year for five years. If you see more than two changes, you’re evaluating a marketing decision. Open their public release notes for the six months after the rename. Count how many entries have anything to do with the new name. When I ran this on one vendor that renamed itself agentic, four of its 53 entries mentioned AI. Two vendors building the capability had AI in most of their recent updates. That gap is the difference between a company that repositioned and a company that renamed.
Ask what changed in the product the month the name changed. Ask for the release notes and read them yourself.
A vendor telling you which category it’s in tells you what its marketing team decided this year. Read the release notes instead. They’re free, they’re dated, and they describe the thing you’re buying.
Methodology: Homepage language captured Sept. 12, 2026, from live pages and the Internet Archive. For one vendor without a single product homepage, I used its DXP product page. I drew release-note analysis from each vendor’s published changelog or release notes. Fifteen of the 17 were measured against dated release notes. Two publish no dated changelog I could reach and were assessed from product documentation, which is weaker evidence. One vendor’s documentation is split across roughly seven products, with no combined feed; I’ve read two. Vendors are deliberately unnamed. Every check described here is reproducible by any reader with a browser.
