We're witnessing something peculiar in the enterprise DXP space. The platforms that once dominated the Gartner Magic Quadrant (Optimizely, Sitecore, Adobe Experience Manager) are locked in a race to out-AI each other. Meanwhile, their core platforms languish with bugs, missing features, and bloated licensing costs. The question isn't whether AI belongs in these platforms. It's whether the current AI arms race is solving problems customers actually have, or merely creating marketing spectacles that distract from fundamental platform deficiencies.
The AI Everything Problem
Let's start with Optimizely and their all-in bet on Opal. Since launching in May 2025, Optimizely has made Opal the centerpiece of virtually every product announcement, investor presentation, and analyst briefing. They've built a team of 30-40 dedicated engineers just for Opal, positioned it as "the industry's first AI-powered DXP," and released benchmark reports claiming dramatic productivity gains.
But here's what they're not talking about: The CMS integration limitations. According to Optimizely's own MVP community members, Opal's "seamless integration" with CMS SaaS is far from complete. Direct integration outside of CMP or in non-SaaS CMS environments remains "still evolving." When your flagship AI assistant doesn't work properly with your own content management system, you have a fundamental architecture problem, not an AI triumph.
The experimentation platform that made Optimizely's name? It's been relegated to a feature that Opal can "enhance." The personalization engine? Now just another context source for AI agents. The content marketing platform? A repository for Opal to mine. Everything has been subsumed into the AI narrative, even when those core products desperately need innovation of their own.
Consider what Optimizely executives are actually saying: CMO Shafqat Islam describes agent workflows that "run autonomously every day or every week" to create experiments, pick winners, and push to production. VP of Product Kevin Li talks about "20X improvements" and "autonomous processes." These aren't incremental enhancements—they're attempting to replace entire workflows with AI.
But our research with actual enterprise customers tells a different story. They want their experimentation platform to have better statistical rigor. They need their CMS to handle governance and compliance more effectively. They're asking for improved integration capabilities and better performance monitoring. Nobody is saying, "What we really need is an AI agent to replace our entire optimization team."
Sitecore's Vaporware Spectacle
Then there's Sitecore, which unveiled "SitecoreAI" at Symposium 2025 with all the fanfare of a product revolution. CEO Eric Stine declared the "AI-first era of digital experience." CPO Roger Connolly demonstrated the new "Agentic Studio" live on stage to thunderous applause. The marketing is slick with promises of "20+ ready-made agents," "no-code agent building," and a "seamless upgrade from XM Cloud."
Dig deeper, and the cracks appear. PathwayAI (their migration acceleration tool) was announced with claims of "70% reduction in migration times" and "nearly 100,000 pages migrated during beta testing." Impressive numbers. But try to get your hands on it as a partner. It's not available. The Agentic Studio that Connolly demonstrated? Partners can't access it to build solutions for clients. These are controlled demos, not production-ready tools.
I've built a career on Sitecore. I respect what they've accomplished. But this feels rushed, like a hasty response to competitive pressure rather than a measured innovation strategy. When you announce major AI capabilities but don't let your partner ecosystem actually use them, you're creating a marketing illusion, not delivering customer value.
The irony is that Sitecore XM Cloud has real strengths that could be amplified. It offers true headless architecture, strong personalization capabilities, and a mature enterprise feature set. But instead of doubling down on making XM Cloud the best headless CMS it can be, they're betting everything on an AI narrative that isn't ready for prime time.
Adobe's Fragmented AI Strategy
Adobe Experience Manager presents a different problem: fragmentation. They have multiple disconnected AI features including AI Assistant for conversational queries, Generate Variations for content creation, and various agents (Experience Production Agent, Content Optimization Agent, Discovery Agent, Development Agent, and Governance Agent) for different workflows. On top of this, they offer Experience Catalyst for content migration, Experience Generation for content assembly, and Sites Optimizer for performance improvements.
It's a constellation of AI features without a coherent strategy. Each product team seems to have added their own AI capability without asking whether it integrates with the others or whether customers need seventeen different AI tools to manage their content.
At Summit 2025, Adobe introduced the Experience Platform Agent Orchestrator to "manage AI agents across Adobe and third-party ecosystems." This is essentially an admission that their AI strategy has become so fragmented that they need a meta-tool to orchestrate all the AI tools. When you need an agent to manage your agents, you've lost the plot.
The fundamental issue is that AEM's core value proposition (unified content, data, and customer journey management) gets diluted when every feature becomes "AI-powered" without clear use cases. Marketers don't wake up thinking, "I need an Experience Production Agent." They think, "I need to publish this campaign across twelve markets by Friday." If AI genuinely helps with that, great. If it's just another layer of complexity, it's a distraction.
The Innovation Theater Problem
Here's the uncomfortable truth: For enterprise DXP vendors, the pressure to appear innovative with AI often comes from internal ambition and competitive anxiety, not from customer demand. PwC's 2025 Customer Experience Survey found that 58% of consumers are only somewhat or not at all comfortable using AI tools to engage with brands, and that "the pressure to implement AI often comes more from internal ambition than from customer demand."
This is innovation theater. The appearance of innovation without the substance. The big DXP players are terrified of being seen as "not AI-native" or "behind the curve" on the technology that's dominating every industry conversation. So they slap "AI-powered" on everything, create elaborate agent frameworks, and commission benchmark studies showing productivity gains.
Meanwhile, actual customer problems go unsolved.
Performance: Customers still complain about slow page loads and bloated JavaScript bundles.
Developer Experience: Complex deployments, unclear documentation, and steep learning curves persist.
Total Cost of Ownership: Licensing costs continue to spiral while feature utilization remains low.
Integration Complexity: Connecting to the rest of the martech stack is still harder than it should be.
Content Governance: Workflow, approval, and compliance features lag behind actual enterprise needs.
These aren't sexy problems. You can't demo them on stage or write press releases about them. But they're what actually matters to the enterprises writing six and seven-figure annual checks.
The Middle Ground: Where's the Balance?
So if the enterprise giants are over-indexing on AI theater, and smaller players like WordPress are (wisely) taking a measured approach with foundational building blocks rather than flashy features, who's getting it right?
Let's look at the headless platforms that are actually solving problems with AI rather than using AI as a marketing distraction.
Sanity: Practical AI Integration
Sanity's Spring 2025 release represents what balanced AI innovation actually looks like. They didn't rebrand their entire platform as "AI-powered." They didn't create an agent orchestration framework. Instead, they added specific, practical AI capabilities that solve real content operations problems:
AI Assist: Contextual content generation and transformation attached to specific fields and documents. Not a chatbot. Not an agent. Just helpful AI exactly where editors need it in their workflow.
Agent Actions: Schema-aware AI instructions for programmatic content operations. These work with your content structure, not against it. They understand your data model and maintain consistency.
Serverless Functions: Event-driven automation that responds to actual content changes. No external orchestration required. No complex workflow tools to configure.
MCP Server Integration: Natural language access to content operations through tools like Claude and Cursor. This leverages existing AI tools rather than building proprietary alternatives.
The key difference? Sanity's AI features augment their core strengths (structured content, real-time collaboration, and developer flexibility) rather than attempting to replace them. The AI makes the platform better at what it already does well. That's the definition of product innovation.
Contentstack: Context Over Hype
Contentstack's approach with Agent OS and their "Context Economy" positioning is more ambitious, but it's grounded in a legitimate insight: AI works better when it understands your brand, your content, and your audience.
Their Brand Kit ensures AI-generated content maintains voice and style consistency. Audience Insights provide the data context that makes personalization meaningful. Polaris, their AI companion, is embedded throughout the platform rather than being a separate tool you have to switch to.
Critically, Contentstack has been transparent about the challenges. CMO Gurdeep Dhillon stated at ContentCon that "AI can do more harm than good if it's not brand aware." CEO Neha Sampat acknowledged that "we can't control the pace of AI and everything we know is about to change. But the opportunity lies in how we adapt."
This is refreshingly honest compared to the "AI will solve everything" messaging from the legacy vendors. Contentstack is building AI capabilities while acknowledging the limitations and risks. That's how you build trust with enterprise customers who have real stakes in getting this right.
What Customers Actually Want
Here's what our research and client conversations consistently reveal about what enterprises actually need from their DXP.
Reliability over novelty: They want platforms that work consistently, have predictable performance, and don't break during routine updates.
Practical automation: They want time-saving automation for genuinely tedious tasks like bulk metadata updates, systematic content transformations, and compliance checking. Not autonomous agents making strategic decisions.
Integration capabilities: They want their DXP to play nicely with their existing martech stack, not be a walled garden that requires proprietary AI agents to function.
Transparent ROI: They want to understand exactly what they're paying for and see measurable returns. "AI-powered" isn't a value proposition if it doesn't solve specific, measurable problems.
Governance and control: They want tools that enhance human decision-making, not replace it. Enterprise content isn't a place to experiment with fully autonomous AI.
Performance at scale: They want platforms that can handle their content volume, user load, and integration complexity without degrading.
Notice what's missing from that list? A desire for agentic AI frameworks. Agent orchestration platforms. Fully autonomous content operations. These are vendor solutions in search of customer problems.
The Path Forward
The irony is that AI genuinely has a role to play in modern DXP platforms. There are legitimate use cases where machine learning and natural language processing can deliver real value.
Intelligent search and discovery using semantic understanding and user intent.
Content recommendations based on performance data and audience signals.
Automated accessibility remediation to ensure compliance at scale.
Smart content tagging and metadata generation to improve discoverability.
Performance optimization suggestions based on real-world usage patterns.
Translation and localization assistance for global content operations.
These are specific, bounded problems where AI capabilities can provide measurable improvements without requiring enterprises to fundamentally rethink their content operations.
But that's not what's happening. Instead, we're getting grand visions of AI-powered content operating systems, autonomous agent workflows, and platforms that "learn as they work." It's ambitious. It's exciting. And it's largely disconnected from what enterprises actually need to publish content, run campaigns, and drive digital experiences at scale.
What This Means for Enterprise Buyers
If you're evaluating DXP platforms in 2025, here's my advice based on two decades in this space.
Look past the AI marketing: Every vendor will claim to be AI-powered. Ask them to demonstrate specific use cases solving problems you actually have. If they can't articulate clear ROI for their AI features, they're selling vaporware.
Evaluate the fundamentals: How's the core platform? Is the CMS actually good at content management? Does the personalization engine deliver meaningful lift? Can developers build efficiently? These matter more than agent orchestration frameworks.
Check partner access: If partners can't get hands-on access to new features, those features aren't ready. Don't accept promises of "coming soon" for capabilities you need today.
Understand the cost model: AI features often come with usage-based pricing, credit systems, or premium tiers. Make sure you understand the total cost before committing.
Demand transparency: How does the AI actually work? What data is it trained on? What guardrails exist? If vendors can't answer these questions clearly, they haven't thought through the implications.
Consider the headless alternatives: Platforms like Sanity and Contentstack are often more honest about their AI capabilities because they're not trying to justify legacy platform complexity with AI magic.
Conclusion: Innovation Without Distraction
The enterprise DXP space is at an inflection point. The incumbents that dominated the last decade are facing legitimate competitive pressure from modern, composable alternatives. AI represents both a genuine opportunity for innovation and a convenient distraction from fundamental platform weaknesses.
The vendors that will succeed are those that use AI to enhance their core value proposition rather than replace it. Those that solve real customer problems rather than creating impressive demos. Those that build trust through transparency rather than hype through marketing.
For now, the big players seem more interested in the spectacle than the substance. They're losing their way not because they lack innovation, but because they're innovating in response to analyst reports and competitive pressure rather than customer needs.
The good news? You don't have to participate in the distraction. You can demand better. You can look past the AI theater to evaluate platforms on their actual capabilities. And you can choose vendors that are building the future thoughtfully rather than breathlessly.
Because in the end, great digital experiences aren't created by autonomous AI agents. They're created by talented teams using powerful, reliable tools that amplify their capabilities rather than replace them. That's what we should be building toward.
And that's what we help our clients achieve at HT Blue—cutting through the noise to find DXP solutions that actually solve problems, deliver ROI, and scale with their business. Not just the platforms with the flashiest AI demos.
