In Box CEO Aaron Levie's vision of the future -- at least, the future of enterprise technology -- everything is instantaneous.

"You are going to have instant, on-demand access to anyone and anything: Whether it's a piece of information, whether it's an asset, whether it's a person that you need to be able to collaborate with," he told Inc. during a recent meeting at the cloud storage and file sharing company's San Francisco office.

In fact, he said, we're already getting a glimpse into that instantaneous future with communications tools like Slack, which give us right-now access to colleagues.

But in coming years, the Box co-founder expects more areas of work to become instant, including the categorization and analysis of companies' proprietary data. And that's where Box, which handles proprietary data, comes in. The platform's users upload roughly one billion files per month and 700 terabytes of data per day, according to the Redwood City-based company.

Levie's hopes for artificial intelligence integration are still a ways into the future for the company. But some application of automation technology is already underway at Box.

For example, a new feature in Box Governance automatically classifies certain documents as confidential based on whether they contain private information such as personal health information, social security numbers, or credit card numbers, according to the company.

One day, with greater AI integration, "the software will be telling you what you should be working on, who you should be communicating with, what information you should be looking at," Levie said.

Box is at the beginning stage of a three-phase process of enhancing its platform with automated features, said Levie. He didn't elaborate on how long the process will take, although the company has indicated in a statement that it thinks "the next decade" will be about software becoming intelligent.

The first phase, which Levie said Box is working on now, can best be summarized as "discoverability." The goal is for Box to automatically categorize data such as images and text documents, so users will be able to search their own data with Google-level ease, he says.

The second phase has to do with recommendations, he said. Say you have two spreadsheets stored by Box, and they happen to contain the same column. A feature might suggest you allow Box to combine the two spreadsheets so you can analyze correlations you are not yet considering.

Or perhaps Box might observe that a large number of colleagues are viewing a certain video uploaded to the platform, and suggest you might like to watch the video as well. The concept of "recommendations" is already familiar to social-media users, Levie noted, who might see trending topics or suggestions on Facebook to wish someone a happy birthday.

Lastly, Levie said the ultimate plan is for Box to be able to perform tasks on behalf of users based on their repeated actions. Does a user often make a folder filled with a certain type of content and add the same people to that folder? In the future, Box might allow for the automation of actions at the click of a button.

Levie said the phases represent a general vision of where things are headed for Box, rather than specific plans that are set in stone. "I'm basically describing my hunch for how you'll see it come together," he says.