California-based cloud content management and file sharing service provider Box recently announced that its Skills Kit platform, which allows organisations and developers to build AI integrations for interacting with stored content on their own, will be available to all customers in December 2018.
The Box Skills framework was first announced in 2017 and was developing an additional layer called ‘Box Skills Kit’ since inception. The latter is a toolkit that allows companies to develop their own bespoke versions of these integrations. The toolkit has attracted development from the likes of IBM, Microsoft, Google, Deloitte, and AIM Consulting.
Chief product officer at Box, Jeetu Patel, said: “Artificial intelligence has the potential to unlock incredible insights, and we are building the world’s best framework, in Box Skills, for bringing that intelligence to enterprise content.”
The Skills Kit has already been used by spirits firm Remy Cointreau. This work involved taking the basic Box Skill for automatically matching metadata to uploaded images, and modifying it so that it would identify specific company products in images. This is how the uploaded images are sorted into specific folders without the need for human interaction or verification.
Box also revealed that its Box Skills platform, which earlier only offered pre-built AI integrations, can now host custom AI models built by third-party AI firms. This means that if an organisation prefers a specific machine learning model built by IBM Watson Studio, Google Cloud AutoML, Microsoft Azure Custom Vision, or AWS SageMaker, can now be integrated into the Box platform to utilise the stored data.
The company also announced updates to its core automation services, which now enables customers to build their own scripts for repetitive workloads. For instance, a marketing team could automate the creation of a template at the beginning of every month and notify specific users to begin collaborating on a new pitch.
Box’s solution appears to be aimed towards smaller work groups that have predictable repetitive tasks in between periods of ad hoc collaboration. It’s less suited for more complicated tasks or those which are unpredictable.
The dashboard for creating these pre-scripted events is very simple, as every automation is based on the premise of ‘if this, then that’. This means that automated processes can be designed quickly by using the drop-down menus.
Box Skills supports over 20 different types of input and output, and includes options for targeting metadata, specific files, or entire folders.
Are you looking forward to the release of Box Skills? Let us know in the comments.