Box Inc. today expanded a beta test program for its Skills software framework that uses machine learning to make video, audio, image and other files more useful on its content management service.

Introduced last year, Box Skills opened up the ability to perform tasks on content such as computer vision for image analysis, video indexing and sentiment analysis from audio using machine learning technologies from IBM Corp.’s Watson, Microsoft Corp’s Azure cloud and Google Cloud.

Now, Box is expanding beyond the approximately 100 customers in the beta program, adding several new customers a week starting in July on top of the likes of Virgin Trains, Ancestry.com, the University of Chicago and the City of San Jose.

“As we see more and more content put into Box, there’s this opportunity with AI and machine learning to have more intelligence put in around the content,” Box Chief Product Officer Jeetu Patel said in an interview.

For instance, among the approximately 600 use cases is a large insurance company building a custom skill to label household objects in images and videos automatically in the homeowner insurance policy process. A construction equipment rental company is building a custom skill to identify equipment and extract part numbers from vehicles appearing in images and videos. Another company is using a Skill to analyze videos automatically with an eye to enabling employees to view only the parts of internal all-hands videos that are relevant to them.

“We’ll try to get it to a few hundred beta customers to start with,” Patel said, depending on feedback it gets. “If it turns out we need to go deeper, we’ll hold off going broader and do that.” He said he expects to expand Skills to a much broader audience in the second half of the year, following the company’s annual Boxworks conference in late August.

Box also announced an expanded relationship with IBM, whose professional services organization will build custom Watson-powered Skills for other companies. In particular, IBM is offering two new services for building the Skills. They included custom image insights that could allow, for example, an environmental organization to analyze satellite images of coastal erosion in Box to detect areas most affected and thus reduced the cost of monitoring, and custom document insights that can automatically tag documents to make finding the right info easier and faster.

“That allows us to scale in a very different way,” Patel said. “It tells you the level of interest we’re seeing in the market, because they wouldn’t do it if they didn’t see the demand.”

Finally, Box is offering support in its Skills Kit for new cognitive services in Azure that Microsoft announced at its Build conference in May. Those include improved optical character recognition for identifying text in images, the ability to detects dozens more objects in an image and tag it, and new speech services for audio analysis, such as the ability to recognize industry-specific terms. Those technologies can be used in both Box and custom Skills.

The upshot of all this is that Box is aiming to position itself as more of a destination for customers to get work done, not just store their data and share it.

“In the next five years, the way that people will engage with content will be very different from the way they have in the past 25 years,” Patel said. “The amount of context and intelligence that can be there around a piece of content will be so much more.”

Image: Box