Boxing clever: How Box is using AI to shake up cloud storage
New machine learning tools take Box beyond a simple storage platform
Machine learning is also set to be threaded through Box's other services. Along with Box Skills, the company also announced Box Graph, a sophisticated system that tracks how people use and interact with content throughout the enterprise, somewhat similar to the way the Google Knowledge Graph catalogues data from the web.
In practise, this enables technology like the new Box Feed, a Graph-powered activity feed that's personalised for every user within the business. This involves highlighting documents that they're collaborating on that may need their attention, trending content that's popular within the wider company and even recommending content based on their activity - much like recommendations engines seen on YouTube or Amazon. Along with Feed, Box Graph is expected to start powering technology like automatic anomaly detection, intelligent searching and more.
Along with the machine learning skills introduced by Box itself, customers will also be able to create their own tools and integrations, thanks to the Box Skills Kit. These developer tools are essentially an SDK for Box Skills, allowing businesses to build their own skills. These can be chained together, can include third-party machine learning systems and can be trained using custom, industry-specific datasets. They will also be compatible with third-party workflow tools in addition to Box's own Relay product.
This is part of Box's efforts to ensure that Box Skills are fully compatible with a broad ecosystem of services and systems. It's also worth noting that the three skills shown off by Box rely on machine learning components developed by IBM, Microsoft and Google. "We're not constrained by how quickly we can build something," Patel said, "we're actually liberally leveraging the entire velocity of the ecosystem and using that as a tailwind."
Open APIs, SDKs and cross-platform compatibility is something that is a huge part of Box's vision for Skills, according to Patel. "We think without that, it won't work. You have to have that. I mean, this is a world where there's going to be massive levels of interoperability," he said; "this is going to be a world that's far more interoperable than a closed system world."
Box's machine learning play adds a huge layer of value to its offering, especially for companies with a large corpus of documents and data. By using machine learning to automatically classify the contents of an organisation's archives, vast amounts of manual grunt work can be eliminated, freeing up time and resources to focus on more productive tasks, Patel explained. "With Box Skills, our goal is to bring intelligence to the content so that the experience that people have with the content in Box is just far more intelligent than what it has been in the past."
"Why Box Skills is important," Lepofsky said, "is because it will help transform box from being just file storage, to being a place where people do actual business processes. If done right, artificial intelligence should never even be seen by the users, it should just add extra functionality that occurs seamlessly."
Customers are excited about the new functionality, too. Michael Duggan, CIO of Oxfam International told IT Pro that the ability to automatically categorise and tag media assets is going to be a huge help to the organisation's outreach efforts.
"Box Skills is incredibly exciting. Oxfam generates a huge amount of content in the field," he said. "The ability to search across that and actually surface content quickly to our public engagement people and our supporters - I think that could be transformational."
This is exactly what Box is aiming to offer its customers with its new machine learning tools. "A piece of content when it's put in Box is infinitely more valuable to the customer than that same piece of content outside of box," Patel said. "The more content that you put in Box, the more value that Box will be able to deliver to you as a customer."