Box CEO Aaron Levie states AI is changing SaaS landscape

Enterprise software is undergoing a structural transformation as artificial intelligence becomes a foundational part of how platforms operate and deliver value. Box CEO Aaron Levie described how AI’s growing role in enterprise workflows is redefining how organizations manage data, automation, and decision-making.

Levie explained that unstructured data - such as documents, emails, contracts, videos, and other forms of content - represents one of the largest untapped resources inside most companies. Historically, software systems have been optimized for structured data in databases, but unstructured information has remained difficult to categorize and act upon. Artificial intelligence now offers a means to analyze, summarize, and automate workflows around this content in real time, unlocking previously inaccessible value.

Unstructured data and evolving workflows

Levie pointed out that automation has traditionally focused on structured workflows handled by systems like customer relationship management, enterprise resource planning, and human resources tools. By contrast, unstructured data workflows have remained largely manual. AI, he said, is now bridging that gap.

Box has developed technology to integrate AI agents directly into enterprise content workflows, giving organizations more control over how automation is deployed. These agents can handle repetitive tasks, process documents, and route information through various business systems, allowing employees to focus on higher-value work.

According to Levie, this marks a shift from isolated applications toward an ecosystem of intelligent agents capable of operating across departments, applications, and data types. The goal is to create seamless collaboration between human workers and digital agents while maintaining oversight, governance, and security.

Guardrails and governance in AI deployment

Levie emphasized that as AI becomes more deeply embedded in enterprise operations, maintaining strong guardrails and governance frameworks is essential. Companies must determine where deterministic rules should apply and where AI’s probabilistic reasoning can be trusted to act autonomously.

He cautioned that giving AI agents too much context or control without limits can lead to inefficiency or unexpected outcomes. Instead, organizations should structure AI deployments into smaller, purpose-built agents - each with defined scopes and permissions. This approach ensures accountability, reduces errors, and simplifies debugging when workflows evolve.

By maintaining control over the decision boundaries of each agent, enterprises can benefit from AI automation without compromising compliance, transparency, or user trust.

Flexibility across AI models and vendors

Levie also discussed how enterprises increasingly prefer flexibility rather than dependence on a single AI vendor or model. Different large language models offer distinct strengths depending on the task - summarization, reasoning, code generation, or analysis—and enterprises want to mix and match these capabilities.

The Box platform is designed to work across multiple AI models, ensuring that customers can choose which technology best fits their data and security requirements. This model-agnostic strategy allows companies to maintain control over sensitive data while taking advantage of ongoing improvements in AI performance and accuracy.

This flexibility also supports global organizations that must comply with diverse data-residency laws and security standards, as they can localize their AI integrations without rearchitecting their entire infrastructure.

Redefining value in enterprise software

Levie said AI’s impact on SaaS is not limited to new features - it is fundamentally changing how software creates value. In the past, applications were built around structured data entry and retrieval. Now, value emerges from how effectively a platform can understand, contextualize, and act on information across multiple systems.

This transition transforms SaaS products from static tools into dynamic orchestration layers that coordinate human and machine intelligence. As a result, enterprises can reimagine workflows such as contract management, marketing campaign analysis, compliance reporting, and product design, all powered by continuous AI interaction.

According to Levie, this shift represents “the era of context,” where success depends on how much relevant data and intelligence a platform can integrate to improve decision-making.
 

The limits and risks of current AI

Despite rapid progress, Levie acknowledged that AI's capabilities still have limits. Large models can become less accurate when given excessive or irrelevant context, and complex workflows can magnify small errors if not monitored carefully.

He stressed that responsible deployment requires transparency, deterministic checkpoints, and well-defined handoffs between agents. Human oversight remains critical, particularly when workflows involve sensitive information, regulatory compliance, or customer communications.

AI systems, he said, should augment human work rather than replace it. When properly supervised, they can significantly increase productivity; when left unchecked, they risk introducing inefficiencies or compliance gaps.

A practical framework for enterprise modernization

Levie’s perspective extends beyond Box itself. He sees AI’s integration into SaaS as a broader shift for the entire industry—one that challenges traditional software architectures. Rather than monolithic platforms trying to solve every problem, the future points toward modular, interoperable systems that allow AI agents to collaborate across services.

This modularity supports faster iteration and easier adaptation to new technologies. It also allows enterprises to incrementally modernize their infrastructure instead of pursuing disruptive, large-scale migrations. For many organizations, that means layering AI capabilities onto existing systems rather than replacing them entirely.

By designing workflows that connect people, processes, and intelligent agents, enterprises can create scalable automation frameworks that evolve alongside their business goals.

The evolving enterprise AI ecosystem

Levie believes the enterprise AI market is entering a maturity phase where interoperability and governance will define competitive advantage. Software providers that prioritize security, auditability, and transparency are likely to earn greater trust from customers managing sensitive information.

He also expects the rise of specialized AI agents for compliance, cybersecurity, and risk management—functions that require strict oversight but can benefit from automation. Over time, AI will help organizations move from reactive responses to predictive insights, identifying risks before they escalate.

This evolution will blur the boundaries between traditional software categories, leading to unified platforms that manage data, automation, and collaboration within a single intelligent ecosystem.

Enterprises embrace a new phase of AI-driven SaaS states Box CEO Aaron Levie 

Aaron Levie’s insights illustrate how artificial intelligence is redefining enterprise software architecture, from how data is processed to how value is delivered. AI is transforming SaaS from isolated applications into interconnected systems of context-aware agents capable of understanding and acting on vast amounts of unstructured information.

The shift requires a balance between innovation and control—leveraging AI to increase efficiency while maintaining compliance, governance, and trust. For enterprises and software providers alike, success in this new landscape depends on flexibility, security, and a willingness to rethink long-held assumptions about what software should do.

As Levie emphasized, the future of SaaS is not about replacing humans with algorithms but about enhancing how people and intelligent systems work together. The convergence of automation, governance, and human creativity is setting the foundation for the next era of enterprise innovation.