Products & Tools

Iron Mountain says only 34% of financial data is AI-ready

Iron Mountain research finds only 34% of financial services content is connected and 96% of firms manually prep data for AI, prompting an InSight DXP update.

Iron Mountain updates platform to aid AI in finance
Iron Mountain updates platform to aid AI in financePexels / pixabay
By Elena Vasquez6 min read

Why it matters

  • Only 34% of financial services content is connected across systems, per Iron Mountain's survey released 18 November 2026.
  • 96% of organisations need manual intervention to prepare unstructured data for AI workflows.
  • InSight DXP claims over 96% extraction accuracy on complex financial instruments, rising above 99% with human review.
  • Vanson Bourne surveyed 400 IT and data leaders across six countries, including 134 financial services and insurance decision-makers.
  • 89% of respondents said legacy content management costs limit AI and modernisation investment.

Only 34% of financial services content is connected across systems, according to new research from Iron Mountain — and 96% of organisations still need manual intervention to prepare unstructured data for AI workflows. The storage and information management company released those figures on 18 November 2026 alongside an update to its InSight DXP platform, built specifically for banks and insurers trying to put AI to work on documents, records and other messy enterprise information.

The numbers describe an operational bottleneck inside financial firms. Staff spend their days classifying, tagging and preparing information scattered across disconnected systems before any AI model can touch it. Among the financial services respondents, 84% reported that highly skilled IT and data science staff get pulled away from strategic work to carry out routine data tasks. Another 45% identified data classification and tagging as a primary manual burden.

The cost side is just as stark. Some 89% of respondents said the cost of legacy enterprise content management limits their ability to invest in AI and modernisation. In other words, the systems firms built to store information are now eating the budget they need to make that information useful.

Varun Gupta, Senior Vice President, Global Product & Solutions Marketing at Iron Mountain, put it bluntly: "Financial services firms are investing heavily in AI, but paying a steep operational tax to feed those models trusted, reliable information."

Why fragmented data blocks AI in finance

The research reflects a wider problem for any company applying AI to unstructured information — the contracts, claims files, loan documents and records that make up most of an enterprise's knowledge. In heavily regulated sectors like banking and insurance, firms face tighter controls over how data is classified, stored and used. That makes AI deployment far more dependent on the quality and governance of the source information than in less regulated industries.

Gupta argues the challenge for most financial institutions is not access to AI tools. It's the quality and organisation of the information underneath them. He said firms need stronger information links and governance if they want to expand AI use beyond isolated pilot projects.

"Connecting that information, preserving its business context and embedding governance into workflows are essential to scaling AI with confidence," Gupta said. "To unlock AI's full potential, financial institutions need a unified content intelligence layer that automates governance and structures dark data at ingestion so models and agents can act on reliable data. In today's AI era, your underlying data foundation is just as critical as the AI technology itself."

That last line is the pitch for the product, but it also captures the industry's current predicament. Generative AI and agentic systems can only act on data they can find, parse and trust — and most financial firms' data fails at the first step.

What does the InSight DXP update actually do?

The latest version of InSight DXP is designed to connect enterprise information and business systems, and it adds automated content intelligence and governance tools. Iron Mountain positions the platform as a way for financial institutions to manage unstructured data, preserve business context and improve oversight across the workflows that feed AI applications.

The update strengthens links between information repositories and operational systems. Its stated aim: cut the manual work involved in turning fragmented content into usable AI-ready data, without breaking the governance rules that regulated firms must follow. The platform attempts to structure and classify content automatically, so that models and AI agents receive reliable, contextualised data rather than raw dark data.

InSight DXP already supports document-heavy financial processes. Iron Mountain lists three core use cases:

  • Trade back office operations
  • Mortgage processing
  • Commercial lending

According to the company, those deployments have achieved more than 96% extraction accuracy on complex financial instruments. When human review is layered on top, accuracy rises above 99%.

Human review stays in the loop

That human-review figure is deliberate. Rather than removing people from decision-making entirely, many firms operating in regulated environments are adding automation to document processing while keeping human checks for higher-risk tasks. Iron Mountain's accuracy claims — 96% automated, 99%-plus with review — are framed around exactly that model: machines do the extraction, people verify the outcomes that matter.

For banks and insurers, this is the pragmatic middle ground. Full autonomy over loan decisions or trade documentation carries regulatory risk. Full manual processing carries the operational tax Gupta describes. The platform update is built for firms trying to land between the two.

Who was surveyed — and what the sector gains show

The research was carried out by Vanson Bourne on behalf of Iron Mountain. It covered 400 IT and data leaders across six countries: the United States, the United Kingdom, Germany, Brazil, India and Australia. Respondents worked at organisations with 1,000 or more employees, spanning banking and financial services, insurance, healthcare and life sciences, technology, manufacturing and the public sector.

A cohort of 134 financial services and insurance decision-makers was analysed separately. That sub-sample produced the sector-specific headline figures: the 34% content connectivity rate, the 96% manual-intervention figure and the 89% reporting that legacy content management costs constrain AI investment.

The sector data is not all bad news. Financial services and insurance organisations reported stronger average gains than their cross-industry peers in the same study. The sector showed 32% higher gains in IT efficiency and 21% higher gains in AI readiness compared with the cross-industry sample. Iron Mountain itself cautioned, though, that these results may mask deeper problems linked to fragmented information management — gains at the surface can coexist with structural data problems underneath.

The stakes for banks and insurers

The commercial logic behind the update is straightforward. Financial institutions are among the biggest spenders on AI, but their returns depend on document-heavy back-office processes — mortgage files, trade documentation, lending paperwork — that remain stubbornly unstructured. Every hour a data scientist spends tagging records is an hour not spent building models, and every disconnected repository is a place where an AI agent cannot look.

Iron Mountain is betting that the bottleneck has shifted. Two years ago, the hard part was getting capable AI models. Now the hard part is feeding them. Gupta's "operational tax" framing targets exactly that shift, and the 89% figure on legacy content management costs suggests plenty of finance IT leaders agree.

Whether a content intelligence layer can actually dissolve that tax is the open question the survey cannot answer. But with 96% of surveyed organisations still preparing unstructured data by hand, Iron Mountain has clearly identified where the pain sits — and the update signals that the next phase of AI competition in financial services will be fought over data foundations, not model access.

Source: itbrief.com.au

Share this article:

More from Elena Vasquez

Elena Vasquez

Show full bio

Market editor covering technology and product news in artificial intelligence and machine learning at AI In Context.

708 articles

Related articles

  1. Only 34% of Financial Services Content Is Connected, Iron Mountain Finds
  2. Iron Mountain: Only 34% of Financial Services Data Is Connected
  3. Novo launches AI banking tools with approval-based controls
  4. Only 10% of Enterprises Let AI Agents Decide Without Human Approval
  5. Gartner: 60% of AI Projects Without AI-Ready Data Will Be Abandoned by 2026

« Previous articleNext article »