In 2025, enterprise AI matured from simple chatbots to agentic systems that can do things like extract insights from unstructured data, intelligently classify that data, and turn it into structured metadata – all while detecting and taking action against security threats. The resulting wave of innovation will fuel innovation at enterprises and smaller companies alike in far more ways than anyone today can imagine.
“AI agents,” says Box CEO Aaron Levie, “will take capabilities that were previously extremely scarce and make them near infinitely available and abundant.” How will this revolution unfold in 2026? We asked Box AI leaders for their suggestions.
Highly trained hero agents unlock maximum value…
Olivia Nottebohm, Chief Operating Officer
As organizations hone their focus on agentic AI, they’re learning that a few strategic agents will outperform many scattered ones. We think that a handful of these “hero agents,” trained and optimized for the right use cases and carefully integrated with corporate datasets, apps, and systems, will yield far more value to any given team or function than many lightly used agents.
…and every agent is one piece of a larger puzzle
Nora Soza, Senior Director of Strategy & Operations
At Box this year, many of our ideas for agents came from thought leaders who are just wrapping their arms around AI and figuring out cool ways to use it. In 2026 we'll keep building these new agents – everyone will – but centrally, we’ll start thinking of them as puzzle pieces that form a bigger picture. A sales team, for instance, will want an agent that knows how to do account research really well, one that does outbound emails really well, agents for deck prep, risk mitigation, market fit. Designing AI agents for specific skills and stringing them together so that the combination brings the bigger picture into focus will deliver our best opportunities for scale
Designing AI agents for specific skills and stringing them together so that the combination brings the bigger picture into focus will deliver our best opportunities for scale.
Developers go from coders to orchestrators…
Tamar Bercovici, VP of Engineering
This year, the use of AI agents for coding has really taken off. In 2026 we’ll see more agents graduate from assistants to teammates, partnering with product and dev teams on the full product development lifecycle, from accelerating ideation and prototyping, to tackling complex coding tasks, code review and release, all the way to helping manage production systems. And the role of developers will evolve as well, from AI adopters to agent orchestrators who drive even more innovation.
…and where developers lead, colleagues follow
Ben Kus, Chief Technology Officer
The success of tools like Codex and Claude Code has made agentic AI a routine part of engineering culture. Today’s developers are spinning up small specialized agents as genuine teammates that manage subtasks, write specs, review code, run tests, and troubleshoot issues in real time. In 2026, that integration of AI into the very fabric of work will spread into other departments across the organization - not to replace people’s roles, but to extend their capabilities.
In 2026, that integration of AI into the very fabric of work will spread into other departments across the organization.
Rigorous security becomes a competitive lever
Heather Ceylan, Chief Information Security Officer
As AI evolves, cybersecurity has to evolve just as quickly. 2026 will bring new challenges in this domain, as shadow AI operating outside established controls creates dangerous blind spots, giving attackers new footholds and forcing enterprises to focus on compliance and governance. Only those organizations that pair the precision of AI-driven defense with strong governance, digital hygiene, and a rigorous approach to security architecture will be resilient enough to withstand the next wave of cyber threats.
Organizational design turns bottlenecks into enablers
Samantha Wessels, Senior VP EMEA GM
Human enablement is still one of the biggest bottlenecks to gaining value from AI. The underlying LLM capability is already there; coaching people to use agents properly is the real frontier. But we won’t overcome the human bottleneck by wrapping chat interfaces around old processes. We’ll clear that hurdle with organizational design – redesigning workflows so that people can focus on judgement, partnership, and customer impact.Lots of companies will start meeting this challenge in 2026, and the happy result will be more human workers, not fewer, becoming highly valuable strategic talent.
Japan will show how AI and humans can truly collaborate
Noriyuki Sato, President K.K. Japan
Here in Japan, because we face the structural challenge of labor shortages due to our declining population, we have a unique opportunity to lead the world in leveraging AI as a new source of workforce capacity. By handling time-consuming tasks like extracting information from documents, automating repetitive processes, and summarizing data for faster decisions, AI opens the door for employees to focus on work that truly requires human skills: creative thinking, strategic judgment, and building relationships. By combining human expertise with AI capabilities, companies can accomplish more without needing to hire more, turning untapped knowledge into real business value. 2026 will mark the dawn of this new era where humans and AI agents truly collaborate.
In other words…
Box leaders see the next generation of agentic tools as smart, specialized, well-trained, carefully guardrailed, and human-led. How will these big-picture themes manifest in actual fine-grained practice? We’ll bring the team back in a quarter or two to take another look.
