Why enterprises need to fix their data before AI breaks their business [Q&A]

There’s been a boom in AI in recent years and the technology has found its way into more and more areas of commercial enterprise.

But in the rush to adopt AI is the quality of the underlying data being ignored? We spoke to Krishna Subramanian, co-founder and COO of Komprise to find out why good data governance is key to implementing AI successfully.

BN: Why are some predicting AI failures this year?

KS: The excitement around generative and agentic AI has led to widespread experimentation but not all of it has been done responsibly. Too often, organizations have plugged these tools into messy, poorly managed datasets, assuming that more data automatically leads to more insight. The reality is that ungoverned data often contains inaccuracies, biases, or sensitive material that can trigger reputational damage, compliance breaches, or outright operational failures.

When failures happen (and they will) they won’t simply be small glitches. They will be high-profile, dramatic, and costly enough to spark a widespread reckoning. These incidents will highlight what many IT leaders already suspect: the biggest risks with AI are not the models themselves, but the data underneath them.

Not only can erratic AI data preparation and ingestion processes derail security and compliance but it can undermine AI monetization. In fact, researchers at MIT found that only around five percent of businesses are generating 'rapid revenue acceleration.'

BN: Does this mean we’re entering an ‘anti-AI’ era?

KS: Not at all. This isn’t a backlash against AI itself, but against careless AI. Enterprises are realizing that reckless adoption without strong guardrails is unsustainable. Rather than scrapping AI investments, CIOs and their teams will be recalibrating: slowing down, inspecting tools more rigorously, and differentiating between solutions that are well-developed, safe, and governed versus those that are immature and risky.

In other words, the hype cycle is cooling, but the technology isn’t going away. What’s ending is the assumption that AI can be bolted onto an enterprise without deep consideration of governance, risk, and data foundations.

BN: What role will data governance play in this recalibration?

KS: AI-specific data governance is about to move from ‘nice to have’ to ‘must have.’ For years, governance frameworks have been designed primarily around compliance, archiving, and retention. But AI introduces new stakes: if data is mislabeled, unclassified, or unmanaged, the AI built on top of it inherits those flaws and magnifies them.

The coming pullback in AI adoption will be less about slowing innovation and more about embedding governance at the core of every initiative. That means CIOs will be implementing systematic auditing, proactive monitoring, and real-time alerts to ensure AI systems aren’t misusing data. It’s not just about regulatory compliance, it’s about protecting business value and preventing catastrophic errors.

BN: Why is unstructured data such a critical piece of the puzzle?

KS: Unstructured data is the elephant in the room for AI adoption. Estimates suggest that 70-90 percent of enterprise data is unstructured: think emails, chat logs, PDFs, design files, videos, or IoT sensor feeds. This data has historically been underutilized because it’s messy and difficult to classify but AI depends on it to provide context and insight.

Without proper classification, tagging, and preparation, unstructured data becomes a liability. Sensitive documents could be exposed to AI models without safeguards, or irrelevant, low-quality data could dilute the accuracy of outputs. IT leaders are now realizing that before they unleash AI, they must first tame the unstructured sprawl, introducing automated tools for tagging, classification, and systematic management.

BN: How can automation support AI data governance?

KS: The sheer scale of data and the real-time speed at which AI consumes and transforms it requires built-in intelligence and automation. Tools that can audit data continuously, detect anomalies, and trigger alerts in real time will form the backbone of modern governance.

Automation ensures consistency, reduces human error, removes bottlenecks, and gives IT teams the confidence that governance isn’t a one-off event but an ongoing safeguard. As AI adoption scales, automation will be the only way to enforce policies at the velocity required.

BN: Beyond technology risk, what larger pressures make AI data governance urgent?

KS: AI doesn’t exist in a vacuum. Geopolitical instability, rising tariffs, and supply chain disruptions are already forcing organizations to think more critically about where their data resides and who has access to it. Data sovereignty is no longer a compliance box to check; it’s a strategic advantage in a volatile world.

By embedding governance into AI initiatives, CIOs can maintain a tighter grip on where data lives, how it moves across borders, and whether it remains secure in different jurisdictions. In turbulent macroeconomic conditions, that knowledge translates directly into resilience. Governance becomes not just a defensive measure, but a competitive differentiator.

BN: How will budget constraints affect AI adoption?

KS: AI is expensive, from procuring the right IT infrastructure (storage and compute) to run LLMs and host AI tools, AI software subscription costs, governance platforms, development costs, and systems and processes to manage and prepare the data. An IBM study indicates that larger companies plan to allocate roughly three percent of their revenue to AI, about $33.2 million annually for a $1 billion company while SMBs may spend five to 20 percent of their total revenue to AI, according to Hubspot.

With these kinds of numbers, CIOs don’t have the luxury of unlimited experimentation. Cost optimization will become a governance issue. The organizations that succeed will be those that can scale AI cost-effectively with technologies that offer flexibility, ease of use and no lock-in. Managing data effectively by seeing trends across data silos where IT can optimize through getting rid of unneeded and duplicate data and storing data on cheaper storage as it becomes less-used are great tactics to cut costs and make room for strategic AI expenses.

BN: What’s the long-term outlook for AI and data governance?

KS: The short-term story is one of evaluation and strategic caution. But the long-term outlook is positive. Organizations that take the time now to embed AI-specific data governance will come out ahead, with systems that are safer, more resilient, and better aligned to business priorities.

Ultimately, governance is not a brake on AI innovation, it’s the foundation for sustainable growth. By taming the chaos of unstructured data, automating oversight, and building cost-optimized, sovereign systems, CIOs can transform AI from a hype-driven gamble into a durable competitive advantage.

Image credit: Khakimullin/depositphotos.com