Managing unstructured data in the era of AI

What’s stopping organizations from understanding 80% of their own data?

Unstructured data, everything from emails and meeting transcripts to PDFs and chat logs, accounts for the majority of enterprise information—as much as 90% of all data, by some estimates. Yet most organizations still struggle to make sense of it.

During Linkurious Days London, we sat down with Aidan Troy, Executive Vice President, EMEA at Nuix, to discuss how businesses can better gain value from their unstructured data. In this interview, he shared practical insights on why managing unstructured data has been so hard for so long, how AI is changing the game, and why off-the-shelf tools are key to staying ahead.

Unstructured data holds valuable insights about how decisions are made, how teams collaborate, and where risks might be hiding, but it’s often locked away in formats that aren’t easy to process. That’s a missed opportunity, especially in fields where speed, accuracy, and context matter, like anti-fraud investigations, intelligence use cases, or law enforcement activities.

As Aidan Troy, Executive Vice President, EMEA at Nuix, points out:

“Many of our customers have been investing for years on managing the 20% of their data that is structured. The unstructured data space is less well invested, less well understood.”

That 80% includes the messy, human side of business: decision-making trails, conversations, approvals, and context that live in formats like emails, Teams calls, PDFs, and Slack messages. And the problem isn’t just scale, it’s complexity and fragmentation.

• It’s everywhere: Unstructured data lives in countless tools and platforms. “I join you on one call on Zoom, I go to another on Teams… the fragmentation of tools is the second big challenge,” Troy notes.
  
• It lacks structure by definition: Unlike rows and columns, unstructured data has no predefined format. That makes it hard to search, store, or query without specialized technology.
  
• No one-size-fits-all analysis: “The insights, the questions, and the interrogation that an organization would like to do with their unstructured data varies enormously, even within an industry,” Troy explains.

AI promises to make unstructured data analysis scalable, but only if paired with the right infrastructure. One of the key technologies driving this promise is natural language processing (NLP).

NLP has opened up new possibilities for understanding unstructured data. It allows machines to parse human language, extract meaning, and detect patterns in formats like emails, PDFs, chat logs, and more.

That means businesses can now search, classify, and make sense of vast amounts of messy, human-generated content, something that would have been too resource-intensive or inconsistent to tackle manually.

But AI alone isn’t enough. Without the right infrastructure, even the most advanced models can fall short. To turn raw text into real insights, organizations need tools that can handle the complexity of unstructured data at scale: storing it securely, processing it efficiently, and ensuring outputs are trustworthy and traceable.

As Aidan Troy points out, many organizations fall into the trap of trying to build their own solutions:

“You can always do the proof of concept using your development teams. making it industrial scale, fit for purpose, so that it doesn't break, and it can always be tracked, that's where the real difficulty is.”

To truly scale AI for unstructured data, businesses need platforms that are built for the job. Here's why purpose-built solutions outperform DIY efforts:

• They’re secure and auditable: With built-in governance, security, and compliance, they reduce operational and regulatory risk.
• They’re ready for production: No need to reinvent the wheel or rebuild prototypes into scalable tools.
• They evolve with AI: As Troy puts it, “Two years ago, nobody had ever heard of ChatGPT… Now, the world is using it.” Platforms must be able to adapt as new models and technologies emerge.

For Nuix, the priority is clear: “We’ve had 20 years building up our expertise and our technology to deal with unstructured data,” says Troy. That includes indexing, categorizing, and making it usable for human investigation and analysis.

But Nuix doesn’t stop at structure, they visualize the meaning inside the mess.

The joint solution, Nuix NLP AI, combines cutting-edge natural language processing with intuitive graph technology to help investigators make sense of complex, connected data. It’s designed to lower the barrier to advanced link analysis, eliminating the need for large teams or niche technical skills. Nuix takes care of ingesting, processing, and enriching unstructured data, while Linkurious brings those relationships to life through interactive graph visualization.

“More and more, people are looking to use visualization as a much more human-centric way of interacting with data,” Troy explains. “We manage the data. Linkurious manages how you build the connections and how you interact with it as a human being.”

The real goal? Combine the 20% of structured data with the 80% of unstructured content, at scale, with speed, and without needing a team of experts.

When organizations fail to connect these dots, “it leaves gaps in understanding, gaps in knowledge, gaps in execution,” Troy says. But with the right tools, those gaps become opportunities.

By combining Nuix’s unstructured data processing with Linkurious’ graph-powered decision intelligence, organizations can go from data chaos to real clarity, and finally make the most of the information they already have.

Want to hear more about how AI and graph technology are transforming unstructured data analysis? Watch the full interview. You can also access the white paper “From raw data to rich insights: Unlocking the power of unstructured data”.