Information overload needs more than AI — think collaboration and automation tech

This is a guest blogpost by John Bates, chief executive officer, the SER Group. In it he talks about why artificial intelligence is not the whole answer to handling vast swathes of corporate information.

Not long ago, office workers routinely carved out time each week to delete, file, and organize emails, documents, and other content. While occasional data tidying still happens, today’s flood of information—amplified by the rise of generative content—means both structured and unstructured data pour into our inboxes and systems nonstop. Increasingly, we depend on AI to shoulder the load: sorting, prioritising, and surfacing what truly matters.

But here’s the catch: AI’s primary role is to generate more information, not manage it. This means AI alone can’t help us make better use of the data we already have. As our reliance on digital grows, global data creation is expected to reach 394 zettabytes by 2028. For example, some University of Chicago scientists say they have recently developed a method to store hundreds of terabytes of data on a small crystal—paving the way for ultra-high-density storage systems that could hold petabytes of data on a single disc.

In the meantime, around 80 to 90% of enterprise data remains unstructured—trapped in PDFs, videos, voice recordings, and images. The irony? PDFs were designed to simplify information sharing, with Adobe still promoting them as tools to ‘streamline workflows, enhance productivity, and maintain a professional image’. Yet, we continue to load these digital files with more data while scheduling countless Zoom and Teams calls to collaborate and push projects forward—only for the recordings to be stored and rarely revisited.

In doing so, we’re adding to what Gartner calls ‘dark data’—valuable business information that remains unused and overlooked. Should we be concerned about the growing zettabyte mountain? Not necessarily, if we shift our perspective from seeing ‘data’ as an abstract entity to recognising it as the core of every business process.

Dark data is the new oil. The opportunity lies in knowing how to mine it and refine it into powerful insight.

Drill baby, drill!

To make smarter business decisions and be exponentially more productive, we need accurate extraction and strategic use of information. While AI will play a central role, it’s the convergence of AI, automation, curation and collaboration techniques that will truly help us turn dark data into business insight. And it’s not just one AI model like ChatGPT; it often needs multiple AI algorithms working together to understand complex documents and other content—extracting meta data, i.e. the ‘meaning’, of the documents, tables, pictures, unstructured text, video and audio. Another step is automating processes around documents, such as paying invoices via ERP systems, matching resumes to job openings, or triggering an order workflow.

Agentic AI adds yet another layer: continuously analysing document archives (Agentic curation) to find opportunities and threats—such as noticing that a key employee is a churn risk because they have been passed over for promotion twice and the tone of their communications is increasingly negative. Lastly, collaboration around content matters too— such as a distributed research team exploring the Natural History Museum’s collection of 80 million artefacts, for example, comparing notes on how climate change has evolved over the last 4.5 billion years.

Such ‘Smart Content’ gives businesses timely access to the right information and allows them to make better, faster decisions at scale. This could lay the foundation for a dynamic corporate knowledge base, where documents no longer become black holes draining productivity. Instead, they transform into active, intelligent assets, not static PDFs or forgotten Teams files.

However, human oversight will remain essential. Transparency, accountability, and sound judgment are critical, and businesses have a responsibility to shareholders, employees, and customers to base decisions on expertise, intuition, and judgment, not solely on the output of an opaque algorithm. In other words, accountability will always be key: Who made the decision? What information was used? How did Frankfurt arrive at that conclusion? The most reliable record will always be a structured, accessible document. So, yes, AI has its role, but it shouldn’t be in control. We need a shift from abstract data management to content that is not only machine-readable but also human-friendly—and ultimately, empowering for humans.

And Smart Content can help flag opportunities and threats (‘Hello! I see an invoice from a trusted partner, but it might be fraudulent based on my cross-referencing with historical patterns; double-checking now’). Thus documents will become key drivers of productivity. With Agentic curation acting as a synthetic knowledge partner, intelligence is embedded at every stage of the content lifecycle—understanding documents and extracting their meaning to automating processes, spotting errors or fraud, and enabling systems to respond with context-aware precision across the enterprise. This is how we move from passive data to active insight.

Business information should seamlessly access, inform, and trigger the right processes in the right systems. If we can achieve this, we wouldn’t need to fear the explosion of digital data. In fact, as zettabyte upon zettabyte continues to accumulate, so too would our ability to unlock and leverage that data for measurable business value.