Most organizations assume their content is ready for AI. Most are wrong. Research shows that half of large enterprises are actively running generative AI in production[i]. Not pilots. Not proofs of concept, AI working at scale inside the business. And yet, in the same breath, 72 percent of those same organizations admit they face foundational challenges with information sprawl and inconsistent content quality.
That contradiction isn’t a minor footnote. It’s the central tension shaping where AI investments succeed or quietly collapse. And it’s exactly what a recent webinar, Activate your content for AI: A practical guide to AI, set out to unpack.
Here’s what three practitioners, spanning research, strategy, and in-the-field implementation, found, and why it matters for the people responsible for making AI actually work inside large organizations.
The hidden cost of unmanaged content
Before you can fix a problem, you have to feel its weight. The Foundry Research sponsored by OpenText, MarketPulse Survey: GenAI Adoption and Readiness, January 2026 puts the stakes in concrete terms:
- 96 percent of senior decision-makers say poorly managed information has caused delays or missed deadlines, and for most, it happens regularly, not occasionally.
- 68 percent say their organization has lost a business opportunity because they couldn’t access information in time.
- Workers at large enterprises spend an average of 2.1 hours every day searching for the information they need to do their jobs.
Those hours don’t show up as a line item on a budget. They don’t generate a ticket or an incident report. They disappear and they compound quietly, until an AI initiative tries to run on that same fragmented foundation and fails.
Three myths that are holding organizations back
Sheila Woo, Senior Director, OpenText Cloud Services, who works directly with enterprises at various stages of AI implementation, identified three myths that consistently point organizations in the wrong direction.
Myth #1: More data means better AI outputs. It doesn’t. More data without curation and governance introduces more noise. AI agents, like the knowledge workers they’re designed to support, need the right data for a specific task, not everything available. Feeding an AI uncontrolled, ungoverned content doesn’t improve its outputs. It dilutes them.
Myth #2: GenAI will fix content chaos. This is perhaps the most consequential misconception of all. AI reflects the environment it’s given. If that environment is inconsistent, fragmented, or poorly governed, AI doesn’t smooth it over; it amplifies the problem. Poor content doesn’t improve when AI touches it. It becomes a bigger risk, faster.
Myth #3: A super AI orchestrator solves the underlying problem. Orchestration is powerful, but it doesn’t replace the need for each agent in the system to have the right, well-governed data context. Think of it like an orchestra: the conductor coordinates, but every musician still needs to know their part. A bad note from one instrument resonates across the entire ensemble. The same is true in an agentic AI pipeline.
Where AI initiatives actually stall
These myths translate into real friction when organizations try to move AI from concept to production. Woo identified three places where progress consistently breaks down.
The first is the gap between having tools and using them effectively. Many organizations have purchased the right systems and written the right policies, but the practice is uneven. Classification is inconsistent across work groups. Governance rules exist on paper but aren’t enforced at scale. This gap becomes highly visible the moment AI is introduced, because AI makes every inconsistency louder.
The second is governance that lives in documents but not in systems. As Mike Safar, Product Marketing Director for AI Content Management, puts it in the session: written policies alone aren’t enough. Governance has to be built into the systems themselves, with guardrails enforced automatically. If users have to remember to apply a policy, many won’t, and no information governance strategy survives that at scale.
The third is content quality issues discovered too late. Redundant, obsolete, or unverified data is a manageable problem before AI touches it. Once AI is introduced, it becomes a direct risk to output accuracy, regulatory compliance, and the trust that leaders need to place in AI-generated decisions.
Intelligent activation: a better way forward
What does getting this right actually look like? The webinar drew on a framework from a whitepaper by Deep Analysis, built around a concept called intelligent activation. It comes down to three focused steps.
Target precisely. You don’t need to activate all your content for AI. Identify the specific collections of unstructured data essential to your priority use cases, and map that data to those cases first. Start with the use case that has a clear business problem and executive backing, then work backwards to the content it actually needs. Don’t boil the ocean.
Understand the data context. Before you connect AI to any content, understand its shape: format, lifecycle, how it was created, how it’s updated, and when it should retire. AI needs to be grounded on the most current, accurate content for the task at hand, not on whatever happens to be in the repository.
Build in protection from day one. Veracity, privacy, and security aren’t post-deployment checks; they’re design requirements. Every AI output should be traceable to a source document. Personally identifiable information (PII) should be masked or access-controlled before AI ever reaches it. And audit trails need to exist not because a regulator asked, but because AI-driven decisions require defensibility, particularly as frameworks like the EU AI Act define more and more business processes as high risk.
These aren’t sequential phases of a long transformation program. They’re the focused, use-case-specific work that separates organizations running AI reliably in production from those cycling through pilots that never scale.
Content readiness is the AI investment
One of the sharpest reframes from the session came in response to a question about making the case to leadership: how do you justify investment in content readiness when the business is already demanding AI results?
Don’t position them as competing. Content readiness is what makes the AI investment viable. You’re not asking for budget instead of AI, you’re asking for the investment required to ensure the AI budget already committed actually delivers.
The organizations getting AI content management right aren’t the ones with the most advanced models or the biggest infrastructure spend. They’re the ones that treated content as the foundation AI runs on, and built governance in before AI arrived, not after something went wrong.
Watch the full webinar to go deeper on each of these areas and how to prioritize content readiness when everything feels urgent.
AI is here. Is your org ready?
See if your enterprise is prepared to turn its AI ambitions into reality. Take the 10-question assessment for your personalized report.
Evaluate your AI readiness
[i] Foundry Research sponsored by OpenText, MarketPulse Survey: GenAI Adoption and Readiness, January 2026
