Box transforms intelligent content management for enterprises

Content powers organizations. Contracts, financial reports, research, presentations, and operational documents form the foundation of how modern enterprises run. Yet for years, much of that information has remained locked in unstructured files—searchable, shareable, but not fully actionable. With Box AI, Box is transforming its secure content management platform into an intelligent system that helps businesses reason with their content, turning stored information into usable insight.

Box is uniquely positioned to help its customers harness AI for content intelligence. “Today we are sitting at an extraordinary time,” says Diego Dugatkin, Chief Product Officer. “The AI revolution is truly transforming how we do things and how things are going to be done in the future.” But, as Dugatkin points out: “To be truly effective, AI requires context.”

For many businesses, that context already exists in the files and documents they use every day, as well as those left unused in storage. “The content that enterprise customers keep in Box is the heart and soul of how they operate their business,” says Dugatkin. Tamar Bercovici, Vice President of Engineering, Core Platform, adds: “We believe that content provides incredibly rich and critical context to enable AI agents to actually do real enterprise workloads and provide value.”

Approximately 90 percent of enterprise content is made up of unstructured data. To help turn that data into tangible business value, the company collaborated closely with AWS. At the core of the solution is Amazon Bedrock, which provides proven AI infrastructure and access to high-performance foundation models—as well as agents through Amazon Bedrock AgentCore.

As Bercovici explains: “Amazon Bedrock is helping us accomplish our AI ambitions by enabling us to connect our customers’ content to the best models out there, while ensuring that our platform meets the security, compliance, availability, and performance requirements that customers expect.” Box was a launch partner for Amazon Nova 2 models in Amazon Bedrock. “Nova 2 Pro models actually showed a 30 percent increase in accuracy relative to previous models,” says Bercovici. “That is exactly the type of technological advancement that we can then translate into higher quality in our product. Our ability to leverage models that combine quality and robustness has been really impactful.”

Through AWS, Box has formed a valuable partnership with Anthropic, finding particular success using Claude models in Amazon Bedrock. “It’s been really interesting to see how strong Sonnet and Opus models have been, particularly in areas that require complex reasoning and orchestration,” says Bercovici. The Box team has found real value in Agent Skills, modular capabilities that extend Claude's functionality. “What Anthropic has provided through Skills is extraordinary in terms of saving us time and helping us bring complex implementations to market faster,” says Dugatkin. “We would have had to develop new skills internally, but instead, we were able to leverage Skills directly.”

Box also leverages the Model Context Protocol (MCP), an open-source standard for enabling AI agents to securely access external systems. Through its MCP integration, Box enables agents in Amazon Quick to retrieve governed enterprise content directly from its platform. “The Amazon Quick integration is another example of how we can extract valuable context from unstructured content to enrich and accelerate business processes, said Bercovici.” As the company continues to expand its AI ecosystem integrations, maintaining enterprise-grade reliability is a constant priority.

By using Amazon Bedrock, Box is able to meet strict service-level agreements (SLAs) and deliver 99.9 percent availability. “Getting the ‘three-nines’ availability guarantee from AWS has been critical to delivering reliability to our customers,” says Bercovici. “Amazon Bedrock has actually enabled us to build quite a sophisticated mechanism of failover between regions, even between model families, so that we can ensure that the Box AI experience is always working for our customers when they need it.”

From an infrastructure perspective, Amazon Bedrock delivers industry-leading security, privacy, and compliance, as well as Amazon Bedrock Guardrails, which help to keep AI outputs accurate and safe. “If you're going to create an enterprise workload that leverages AI, you need to also trust that your data is secure,” says Bercovici. “It’s one of the best things about working with AWS, it's a platform that we can trust, and that our customers can trust too.”

Beyond technology, Box is also collaborating with AWS on commercial initiatives. For example, the Box platform is available on AWS Marketplace, helping to further streamline procurement of Box AI for Amazon Bedrock users. “AWS Marketplace enables us to tell a joint story about the value that we can provide to our customers,” says Bercovici. “It’s a fantastic way to make Box products more easily available to our joint customers, and it also simplifies the integration between those products and the broader AWS ecosystem.”

Box recently launched the Box Agent, an AI capability that can turn content into context. Leveraging powerful AI tools from AWS and Anthropic, the Box Agent takes natural-language prompts to execute enterprise tasks end-to-end, transforming how businesses can extract value from their content. “Box has evolved from a place where you securely store content in the cloud, make it governable, and enable collaboration, to a platform where you can truly extract intelligence from all your unstructured data and make it actionable,” says Dugatkin. “Amazon Bedrock has been instrumental in accelerating our roadmap and guaranteeing that we continue to increase the reliability of the overall Box platform.”

Going forward, Box will continue to collaborate with AWS to deliver new capabilities to its customers. “It’s been really great to work with a partner like AWS who, similarly to Box, believes in customer choice,” says Bercovici. “AWS has been such a fantastic partner for our infrastructure over many years, and as we all embark on this AI journey, it's been really great to be able to use AWS and Amazon Bedrock.”

The Box team anticipates rapid progress in the field of agentic AI in the near future. “In the next five years, I see agents truly transforming the way we work,” says Dugatkin. For Box, strategic collaboration with AWS will play a significant role in shaping that future. “We believe that there is no winner takes all, and the solutions will come from these very integrated partnerships that combine the best that each one has to provide.”