TLDR: AI agents are transforming how enterprises manage and derive value from their vast stores of unstructured data, such as PDFs, videos, and images. By automating classification, enhancing data retrieval through techniques like Retrieval-Augmented Generation (RAG), and evolving into sophisticated collaborators, these agents are unlocking critical insights and driving more intelligent workflows, according to Box CTO Ben Kus at Okta’s Oktane AI event.
The enterprise landscape is undergoing a significant transformation as artificial intelligence agents emerge as a pivotal technology for unlocking value from the deluge of unstructured data. This shift was a key topic at Okta’s Oktane AI event, where Ben Kus, Chief Technology Officer of Box Inc., highlighted how AI agents are enabling organizations to convert static files into dynamic sources of decision-making power. The discussion was part of an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio, on September 26, 2025.
Unstructured data, encompassing everything from PDFs and videos to images and audio files, now constitutes the overwhelming majority of organizational content. Historically, companies have grappled with the challenge of extracting meaningful insights from this hidden trove, often finding classification to be a costly and error-prone endeavor, despite its necessity for security and compliance.
According to Kus, AI agents are now turning the tide by automating the classification process at scale. This allows for the efficient tagging of sensitive information and the identification of high-value content across millions of documents. Beyond mere classification, a second foundational step involves Retrieval-Augmented Generation (RAG). This advanced technique empowers AI systems to retrieve precise answers directly, rather than simply pointing to relevant documents. By integrating RAG with enterprise content, businesses can rapidly access critical information like case studies, customer histories, or HR data that would otherwise demand extensive manual searching.
Box, a company deeply rooted in managing unstructured data, sees its role evolving to ‘secure it, to collaborate with it and then now to be able to provide AI on top of your unstructured data,’ Kus stated. He emphasized that AI agents are moving beyond basic bots to become sophisticated collaborators, driving more intelligent workflows and personalized recommendations.
The next wave of enterprise AI, Kus explained, hinges on how effectively businesses can harness this vast pool of information. This necessitates a new technological layer capable of ‘reading, classifying and reasoning over content the way humans do.’ For enterprises adopting these generative AI solutions, governance and security remain paramount concerns, a sentiment echoed in a recent Gartner survey that identified it as the number one issue.
Also Read:
- SAS Unveils Retrieval Agent Manager to Enhance Generative and Agentic AI Capabilities for Enterprises
- New Cross App Access Protocol Fortifies Enterprise AI Agent Security
The evolution of AI agents is poised to unlock substantial value, potentially contributing to a multi-trillion-dollar market. As these systems become more adept at planning, reasoning, and executing tasks by leveraging existing generative AI models, the focus for enterprises will be on building robust governance and security frameworks to manage this powerful new capability effectively.
