Relying on file storage heritage, Box pivots to AI

AI agents will fundamentally change the value of content and the way people work with data and files, said Aaron Levie, CEO of Box, during a Wednesday webcast that was part of the company’s Content+AI Virtual Summit.

The traditional approach to managing content is fundamentally changing with AI, Levie said.

Over the last two decades, Box has changed its focus from file storage to extracting value from content in those files. The company provides collaborative tools and integrates popular apps for users to work with those files and data.

Extracting value from the content in stored files will take on new meaning with AI. The amount of unstructured data within files has grown at an exponential pace over the last few decades, but most of the data is underutilized, Levie said.

AI will provide instant answers from unstructured data, fundamentally changing the value of content stored in systems. It will also automate workflows and introduce an AI-first culture, Levie said.

“We have the opportunity to drive an incredible amount of new experiences,” Levie said. “We need a better and modern way to manage information.”

Box has kept up with various stages in AI evolution, with integration of AI models and more recently AI agents. As AI advances, the vendor is allowing its clients to do more advanced data extraction, multi-step reasoning, and more complex task planning.

“We didn’t think about AI as an add-on capability on the side,” but as central to the Box platform, Levie said.

For example, the Box interface has an agent that can answer queries and work with content within specific folders. It provides multimedia responses to queries, linking to videos, charts, and citations within files stored in the system.

The company on Wednesday launched a series of new AI agents that serve different objectives, including a “Search” agent for basic results from files and a “Deep Research” agent that digs deeper into information for more comprehensive answers.

The Deep Research agent works with large volumes of enterprise content in files, summarizes findings, and provides links to relevant files.

Box is also integrating a new AI agent for Microsoft 365 Copilot, allowing users of Word, PowerPoint, and other software to work with data stored in Box systems.

In the coming years, Microsoft plans to integrate thousands of agents from third parties such as Box and Adobe to improve user productivity.

Many of Box’s competitors are also offering their own AI technology, though agent-to-agent integrations between vendors are growing.

Forrester Research has a roadmap for broad adoption of various agent technologies in coming years. It shows helper-style agents like the ones announced by Box taking off gradually in 2026.

Helper-style AI agents will see broad adoption by next year, according to Forrester. More complex agents capable of executive decision-making will take a little longer.
Forrester Research

But as AI agents advance to solving problems and making executive decisions, things will get more complex, said Craig Le Clair, vice president and principal analyst at Forrester.

There’s a huge gap between vendors of AI agents and the actual adoption profile of users and buyers, Le Clair said. “The gap that exists there is astounding,” he said.

Companies are advancing with AI agents, but it’s much more complex when it comes to sophisticated solvers and managing agents, Le Clair said.

“If you’re in a financial institution, you’ve built all these layers of process, control, risk mitigation, and reporting around the technology underneath. And you don’t change that easily, because it takes legal, compliance, meetings, security. It takes longer to change the processes that sit above the technology,” Le Clair said.