AI is a capability expander. Its role is not merely to act as a shortcut to getting the usual work done, but to enable new ideas and workflows you’ve never thought of before.
With AI agents, you can now create automated workflows that span systems and connect datasets you’ve always assumed couldn’t be connected. You can redesign the way you do things to focus not just on efficiency and assistance, but game-changing progress. Most importantly, you can hold your people up to their highest potential, enabling them to act on fresh ideas and foster their curiosity to inspire real enterprise innovation.
When we say “you,” we also mean us, Box.
This fundamental philosophy of AI as a capability expander drives so much of what we do at Box, not just in terms of how we build products, but in how we leverage AI for our own work behind the scenes. Box is on a journey to become an AI-first company, and that means fostering a culture of AI curiosity and enablement across all our teams and workflows.
But AI as a “capability expander” is just the first of the five ways we think about AI at Box.
Key takeaways:
- At Box, 5 principles anchor our AI-first approach in integrity, including always keeping humans in the loop and prioritizing content security
- By leaning into the capabilities of AI, we’ve saved Boxers hundreds of hours on answering customer questions, preparing for sales meetings, and creating SOWs
- As we move along the AI maturity curve, AI will accelerate decision-making, help surface relevant content, and amplify employee productivity — not replace human work
The 5 principles that anchor our AI-first approach
Thinking of AI as the key to true innovation is number-one to our AI philosophy at Box. But four other principles anchor our AI-first approach in integrity and help us focus on applying AI in the right places and the right ways. Altogether, the five principles are:
- AI as a capability expander: As we mentioned, the ultimate potential of AI is to accelerate and transform work in ways that have not been possible until now.
- Human-AI partnership:If there’s AI involved, there’s always a human in the loop. Humans still ultimately make decisions and issue approvals.
- AI-native design: We start with a clean slate and design from scratch, knowing what AI is capable of — and how it will inevitably evolve.
- Privacy, security, trust, and governance: We always ensure we’re deploying and using AI in a safe and secure way. That’s table stakes.
- Data as a strategic asset: Unstructured data, in particular, powers the way we work.
These are the principles we consider when applying AI to how we work at Box. Luckily, we already have all our content in one place, and that sets us up to take full advantage of Box AI tools and capabilities, including AI Agents.
Thinking of AI as the key to true innovation is number-one to our AI philosophy at Box.
Here are a few key examples of how we use Box at Box — which we call “Box on Box.”
Box on Box for better customer support
At Box, our customer success team was fielding over 10,000 customer support questions a year across many channels. Obviously, we want to answer all these questions quickly and correctly. Our customer success managers were spending a lot of time manually sorting through product articles, old chats, and emails to try to find the best answers — even when the questions had been answered before.
We knew these high-value team members could better spend their time solving new problems. Of course, you can’t anticipate every possible question a customer might ask, so as most customer service organizations will tell you, it’s never as simple as creating a FAQ for reps to respond from. We needed a more sophisticated way to surface the right answers to specific questions quickly, every time, from answers lodged in different systems and types of content.
How we solved this: We already had the content — thousands of internal customer support articles that contain the answers to every conceivable question a customer might ask. We built a Box Hub for customer support, with all of the articles pulled into that Hub via an API. We then combined those articles with other existing product content, and applied a Box AI Agent to make them queryable. Finally, we added some contextual user instructions.
In turn, our customers get clear, reliable answers to their questions in seconds. More than half of support questions are now answered in a self-service way — saving Boxers over 1,000 hours to date. This makes for better customer support, yes, but also a more efficient sales organization.
More than half of support questions are now answered in a self-service way — saving Boxers over 1,000 hours to date.
Box on Box for a stronger sales organization
Second use case: sales. Like in most organizations, our sales reps traditionally spent a lot of time researching customers, building outreach, and prepping for key meetings. It’s high effort, fairly repetitive work that can easily be done by AI.
How we solved this: We re-architected the sales process with the help of Box AI by creating a series of purpose-built agents that now streamline every step of the sales cycle so our reps can focus on the human tasks that drive results. A research-focused AI agent delves into the business goals and challenges of our prospects and reports back on how they align with Box capabilities. We also have a messaging agent that builds outreach based on that research — always with human approval as part of the workflow.
So far, each sales rep and customer success manager has saved about five hours a week.
Box on Box for more streamlined professional services
And finally, we tackled SOW creation — a critical step in landing new customers. We already had created more than 1,100 SOWs, but manual processes made it hard to reuse and reference them. Every new SOW was a from-scratch project, at best taking advantage of copy and paste.
How we solved this: AI was made for this. We added knowledge to the brand-new product Box Extract to capture key metadata for fields in our SOWs — clause language, scope, deliverables, etc. — and extracted relevant data from the most successful existing SOWS. Next, we use Box Doc Gen to assemble new SOWs, sending them out for customer signature with Box Sign. Abiding by our own AI philosophy, a “human in the loop” always checks to make sure that the SOW makes sense before it’s sent.
As a result, we’ve reduced SOW creation time by 35% — and our teams have gotten 2,000 hours back so far.
Hustling along the AI maturity curve
These three examples also illustrate how we are navigating the AI maturity curve here at Box. Enterprise use of AI usually matures in a pattern that looks something like this:
- AI is primarily used for basic information retrieval: Query AI, get a dependable answer thanks to basic retrieval — but humans still make all decisions and execute all workflows
- AI executes simple workflows across departments: Use AI to complete a task on your behalf, with a predefined workflow and human oversight
- AI systems interact with each other to solve complex problems: Run an entire end-to-end enterprise process using agents within a workflow, with humans overseeing decisions and setting the strategic vision
- AI self-improves: With limited human intervention, AI agents improve upon their behavior and self-correct — humans mainly approve strategic changes
- AI independently manages entire functions: At the very highest level, AI could manage entire business functions, identifying needs and self-correcting on their own
The customer success use case mentioned above falls into the basic retrieval and insights stage — a foundational use of AI here at Box. The sales use case, though, begins to progress Box along the maturity curve toward executing workflows. And with the professional services use case, you begin to see how Box is entering the third stage of AI maturity.
As Box (and your organization) moves along the maturity curve, AI begins to self-improve within limited human intervention. But even at the very height of AI functioning, humans are always in the loop. The goal is for AI to accelerate decision-making, help surface relevant content, and amplify employee productivity — not replace human work. So even at stage 5, humans oversee strategy and are the ultimate decision-makers.
Even at the very height of AI functioning, humans are always in the loop.
In the AI-human equation, we want our Boxers to be able to focus on elevated thinking, innovation, and relationship-building. Data-driven tasks and workflows? Whether routine or complex, those are for AI.
Ready to become an AI-first enterprise leader?
To learn more about how Box is innovating and evolving with AI — and actionable steps your organization can take — download our white paper: Becoming an AI-First Company, and check out a conversation with Box CEO Aaron Levie about building an AI-first strategy.
