Three Pillars of an AI-First Organization
Box’s own “Box on Box” journey points past individual productivity: redesign work with AI in mind, give agents clean governed data, and bring every employee along—not leave them to nights and weekends.
Companies that sell transformation advice look more credible when they transform themselves. Box’s internal journey—sometimes framed as “Box on Box”—yielded three lessons that travel well beyond any single vendor: how you design work, how you treat data, and how you treat people.
Design the work with AI in mind
Individual productivity matters—faster drafts, quicker research, fewer blank pages. The larger unlock, though, comes when enterprise-level processes are redesigned with AI at the center. Bolting a chatbot onto yesterday’s workflow is like adding a motor to a horse cart: motion improves, the route does not. Real impact shows up when the process itself assumes what agents and models can do.
Agents need clean, permissioned data
Everyone says “data matters.” It is still true in a sharper way for agents: they need content that is clean, curated, permissioned, and governed. Without that, automation amplifies noise and risk. Content platforms sit in the middle of that problem by design—which is why the lesson doubles as a product story for Box—but the operating point stands on its own. If the knowledge an agent touches is a mess, the agent will be a confident mess.
Bring everyone with you
The third pillar is change management—human behavior. Box’s internal phrasing is blunt: no Boxer left behind. The stance is not “figure it out on nights and weekends or you are off the boat.” The enterprise owns the duty to educate people so they are ready, familiar, and leaned into AI. Leave adoption to the already curious, and you create a two-speed company: a few power users and a quiet majority that never trusts the tools.
- Process: redesign work assuming AI, not decorating old steps.
- Data: clean, curated, permissioned, governed—or agents cannot be trusted.
- People: education and inclusion are operating duties, not side hobbies.
None of the three pillars works alone. Clever process design on messy data fails. Great data with no one trained fails. Broad training on yesterday’s workflows underuses the technology. AI-first is less a slogan than a three-legged stool: tip any leg, and the whole thing wobbles.