Box, the enterprise cloud content management platform, just held their annual developer conference, Boxworks. At our annual user conference, CEO Aaron Levie announced a brand new, industry-leading suite of artificial intelligence (AI) features. The annual developer conference, held in San Francisco, served as a showcase for Box’s new focus on bringing agentic AI models into the center of its core product. Levie said security procedures and access controls is critical in any AI deployment, especially with unstructured data like communications.
At Boxworks, Levie explained that Box has invested nearly two decades in developing a robust system for managing data access across enterprises. This security system allows access only to those with permission. This element is especially important because AI is changing the workplace fast and furiously. Levie put his finger on a big problem with AI deployments. All users fall into a trap of thinking that unlimited access to unfiltered, unstructured data means accurate results.
“It’s where a lot of AI deployments go wrong. People think, ‘Hey, this is easy. I’ll give an AI model access to all of my unstructured data, and it’ll answer questions for people.’ And then it starts to give you answers on data that you don’t have access to or you shouldn’t have access to,” said Levie during his discussion with TechCrunch.
Levie’s concern on the changing nature of work as a result of AI illustrates a bigger trend among tech leaders today. He noted that automation has advanced a lot in processing structured data, particularly around CRM and ERP. Unstructured data is every bit as critical, and it continues to go mostly underutilized. Legal reviews, marketing asset management, and M&A deal assessments often include large amounts of unstructured data. This moment presents tremendous challenges and opportunity for responsible AI integration.
“So the thing that we think about all day long – and what our focus is at Box – is how much work is changing due to AI. And the vast majority of the impact right now is on workflows involving unstructured data,” Levie explained.
Box’s innovative approach builds upon what Box already has in place to improve capabilities with AI. By inferring context from unstructured data, Box is focused on setting the stage to make AI agents more effective. All that considered, Levie thinks context is the key to AI working successfully.
“I think we’re in the era of context within AI. What AI models and agents need is context, and the context that they need to work off is sitting inside your unstructured data,” he stated.
The CEO pointed out the limitations of today’s generative AI systems. He particularly highlighted their difficulty on tasks requiring extended contextual understanding. He stressed the value of dividing workflows into bite-sized pieces to help maintain uniform decision-making from AI agents.
“We’ve already seen some of the limitations even in the most advanced fully agentic systems like Claude Code. At some point in the task, the model runs out of context-window room to continue making good decisions,” Levie remarked.
Box’s advancements are timely, especially as other technology companies like Anthropic introduce new features related to file uploads for AI models. This competitive landscape further highlights the need for businesses to identify and implement secure, scalable AI solutions.
Levie underscored the urgent need for organizations to build security into their systems as they roll out AI at scale. He’s convinced that organizations should have full control over the permissions and access level of their users. Alongside this, they need to have the flexibility to select the AI models that best meet their dynamic needs.
“So if you think about what enterprises need when they deploy AI at scale, they need security, permissions and control. They need the user interface, they need powerful APIs,” he said.
The launch of new AI features during Boxworks signifies a pivotal moment for both Box and its clients as they navigate the complexities of integrating AI into everyday workflows. Levie’s recent comments underscore the company’s focus on developing such a framework in order to drive transformational productivity improvements. Simultaneously, it works to ensure that sensitive information remains protected.
