HyperAIHyperAI

Command Palette

Search for a command to run...

Aaron Levie: AI Agents Won’t Replace SaaS, But Will Transform It — Enterprise Software Must Evolve to Support Agent-First Workflows and New Consumption Models

Box co-founder and CEO Aaron Levie believes AI agents won’t replace enterprise SaaS companies, but instead will work alongside them in a hybrid model. Speaking at the TechCrunch Disrupt 2025 conference, Levie described a future where traditional SaaS platforms continue to serve as the backbone of mission-critical business processes, while AI agents operate on top to enhance productivity and decision-making. He emphasized that deterministic systems—those with predictable, rule-based logic—are essential for core workflows. “Once you have a business process, you want to define it in business logic with systems that are reliable and consistent,” Levie said. “The risk of something changing unexpectedly is too high, especially when it comes to things like data security or production systems.” He pointed to real-world examples where agents have caused data leaks or unintended system changes, underscoring the need for a clear separation between reliable, rule-based software and the more unpredictable nature of AI agents. In this vision, SaaS platforms will handle the core operations—like document management, compliance, and workflow execution—while AI agents act as intelligent assistants, automating tasks, offering recommendations, and speeding up user actions within the system. “The agent is not replacing the system—it’s helping the user get more value from it,” Levie explained. This shift, he noted, will also transform the business model of enterprise software. With AI agents expected to number 100 to 1,000 times more than human users, the traditional per-seat licensing model becomes obsolete. Instead, companies will need to adopt consumption-based pricing models that scale with agent usage and data volume. Levie believes this transition creates a major opportunity for startups. Unlike large, established SaaS companies that must retrofit AI into existing processes, smaller players have a clean slate. “They don’t have legacy workflows to change. They can build from the ground up with an agent-first mindset,” he said. This allows them to design more efficient, scalable, and intuitive systems tailored to the new era of AI-driven work. He encouraged entrepreneurs to seize this moment, calling it a rare and pivotal window in tech history. “We’re in a platform shift unlike anything we’ve seen in about 15 years,” Levie said. “A new generation of companies is emerging, and the time to build is now. You should fully exploit this opportunity.”

Related Links