Introduction
Box is a Redwood City, California-based public company that built one of the earliest cloud file storage and sharing platforms for businesses before evolving into what it now calls an Intelligent Content Management, or ICM, provider. Founded in 2005 by Aaron Levie and Dylan Smith, the company has spent the past two years repositioning its platform specifically as infrastructure for AI agents rather than only for human collaboration.
What the company does
Box’s core platform lets organizations store, secure, and collaborate on files in an online system accessible from any device, with tools for commenting, sharing, applying workflow automation, and enforcing security and governance policies over content. The company’s more recent Box AI layer natively integrates large language models from providers including OpenAI, Anthropic, and Google directly into that content platform, letting users generate summaries, extract structured data from documents such as contracts and bank statements, and receive AI-powered recommendations without leaving Box.
Who it serves
Box sells to enterprises across content-intensive and often regulated industries, including financial services, healthcare, legal, and media, that manage large volumes of unstructured data such as contracts, proprietary documents, and financial records — data the company says accounts for roughly 90% of all information inside a typical organization. These customers are typically drawn to Box for its combination of AI-powered content tools and enterprise-grade security, compliance, and governance controls rather than storage capacity alone.
Company background
Aaron Levie, then a University of Southern California business student, began developing the idea for what became Box in 2003 and 2004, dropping out of school in 2005 to build the service full time with longtime friend and co-founder Dylan Smith, working first in the attic of Smith’s parents’ house and later in a converted garage. Early co-founders Jeff Queisser and Sam Ghods joined shortly after. The founders relied on personal savings and support from friends and family before entrepreneur Mark Cuban invested $350,000 in seed funding in 2005. Box shifted its focus from individual consumers to business customers around 2009 and 2010, raised roughly $500 million across multiple funding rounds, and went public on the New York Stock Exchange under the ticker BOX in 2015. Levie continues to serve as CEO and Smith as CFO, both more than two decades after founding the company.
Product and AI capabilities
Box AI lets users tap into leading third-party AI models for real-time summaries, deep document analysis, and smart recommendations directly inside the content platform, while Box AI Studio, aimed at administrators rather than end users, provides a no-code interface for building, testing, and deploying custom AI agents. Those agents can be configured with specific instructions, scoped to particular knowledge sources such as individual files or content hubs, and assigned a chosen underlying AI model, then tested in a secure playground before deployment to specific users. Box has also emphasized interoperability over building a single proprietary agent: the company has publicly demonstrated its content platform working with OpenAI’s Codex, and says it integrates with Anthropic’s Claude Cowork, Cursor, Microsoft Copilot, Google Gemini, IBM watsonx, and Salesforce Agentforce, positioning Box as the secure, governed content layer that a customer’s chosen agent — whichever vendor built it — connects to.
Key developments
Box’s fiscal year 2026 Form 10-K describes the company as the leading Intelligent Content Management provider, formalizing a positioning shift away from its earlier identity as a cloud storage and file-sharing vendor. The company has continued to expand Box AI Studio’s agent-building capabilities and has publicized its interoperability with a widening set of third-party agent platforms, including a 2026 demonstration of Box working with OpenAI’s Codex to bring AI to knowledge work in a headless, API-driven fashion. CEO Aaron Levie has been particularly vocal about what he describes as the enterprise context problem facing AI agents — the challenge of connecting agents to the large amount of unstructured content that lives in systems like Box, rather than only to structured databases.
Why it matters
Box’s bet that it can succeed as the secure content layer underneath many different vendors’ AI agents, rather than by building the dominant agent itself, is a distinctive strategy among the companies in this batch. If AI agents increasingly need governed, permissioned access to a company’s unstructured content to be useful, Box’s two-decade head start on enterprise content security and compliance could prove more durable than any single agent’s individual capabilities — a bet the company is actively testing through its growing list of agent-platform integrations.
Sector context
Within enterprise software, Box’s interoperability-first strategy contrasts with vendors like Asana and monday.com, which are building and marketing their own named AI agents, positioning Box instead as neutral infrastructure that any agent, regardless of vendor, can rely on for secure content access. That positioning also puts Box in a complementary rather than directly competitive relationship with workplace search tools such as Glean, since both companies are solving related but distinct problems: permissioned retrieval across systems for Glean, and secure, agent-accessible content storage for Box.
Sources and references
This profile draws on Box’s official website and its fiscal year 2026 Form 10-K filed with the SEC, cross-checked against Wikipedia’s corporate history entry and Plunkett Research’s company profile for founding and leadership details. Readers should consult Box’s SEC filings for the most current financial figures.