Computerworld

Box puts automation at the heart of its product vision

Vendor increasingly veering into the automation space thanks to machine learning
  • Scott Carey (Computerworld UK)
  • 17 October, 2018 05:00
Jeetu Patel (Box)

Jeetu Patel (Box)

Enterprise cloud storage vendor Box is increasingly looking to help its customers automate repetitive or mundane business tasks.

Speaking to Computerworld UK during a UK visit, Jeetu Patel, chief product officer and chief strategy officer at Box, explained how the company's product roadmap aligns with three goals.

"If you think of the higher order problems that we are solving right now there are three categories," Patel explained.

"The first one is helping our customers with building a better digital workplace, so giving employees the tooling to be more productive, transparent and collaborative with each other so their organisation can operate in a more efficient and agile manner and innovate faster.

"That's probably the one we have been solving for the longest amount of time, that's the core of what we do.

"The second area is what we call digital business, so helping the company itself operate as a digital company rather than an analogue company.

"So everything from how they engage with customers to business processes and supplier and partner networks and how they engage with those people and create a level of agility in those external engagements and automate business processes.

"That's probably the newest venture we have gotten in to over the past three years and we have been investing pretty heavily in that area.

"The third is security, data protection, privacy and compliance to make sure we can continue to keep innovating there in a meaningful way."

So while Box is still a content management and collaboration specialist at its core, it is steadily moving more into building a layer of what is called robotic process automation (RPA) software to help its customers manage these very manual business processes.

Building an automations engine

The vendor announced a range of product news at BoxWorks in San Francisco earlier this year which covered all three of these areas, from core product enhancements like activity feed and recommended apps, to new security and compliance features, including two-factor authentication for third-party collaborators.

Like every other enterprise software vendor worth its salt, Box is exploring machine learning techniques to improve its core product, and clearly sees the technology's ability to automate mundane processes as the most valuable application for customers.

The result is what Patel calls "an automation layer" which is built on a simple action/event framework, so when a certain event occurs regarding a piece of content within Box, that triggers a resulting action.

"So if a file gets uploaded to this folder, it will assign a task to this team member. Those kind of single task automations can be hugely useful in a multitude of different tasks," he explained. These capabilities will start to be rolled into the base Box product early next year.

He uses an example tailored for me to in order to illustrate this.

"This article you may write, you might want to share it with a couple of colleagues and send to your editor and when that happens it is all manually done in an email, so we wanted to go after that workload and make it lightweight and solve that problem," he said.

The result is essentially an automated copy approval process, automating the handoff of a project from the copywriter to the client and to legal, without having to attach it to an email every time.

It's a simple use case with huge potential productivity gains, not just for media organisations, but for marketing teams and legal practices around the globe.

By being a neutral platform, Box can tie that process into other SaaS apps like Salesforce CRM or Adobe Sign, making these handoffs seamless thanks to API connections.

"That all happens manually right now, so with automations you can create those actions within the system," Patel added.

Box skills

The work the vendor has been doing with Box Skills is a key example of this subtle shift into workplace automation.

The machine learning powered skills, and the open Box Skills Kit for more bespoke models, both allow customers to build more automated workflows around their content.

"Box Skills is a first-of-its-kind framework that will make it possible to digitise almost any business process on Box," CEO Aaron Levie said at the time of release.

The initial Box Skills focused on Audio Intelligence, which uses IBM's Watson technology to show how audio files can be used to create and index a text transcript, which can then be searched; Video Intelligence using Microsoft's Cognitive Services to provide transcription, topic and people detection; and Image Intelligence, which uses Google Cloud to detect objects and concepts in images, capturing text using optical character recognition (OCR) and automatically adding labels to images to build metadata on image catalogues.

The difference between Skills and Skills Kit can be illustrated by the example of a white trainer. The basic Box Skill can tag an image of a white trainer with 'shoe', 'white', 'trainer' etc. whereas a custom-built Box Skill from Nike could identify that shoe as a 1997 Air Jordan XII.

Data entry, digital asset management and metadata tagging are all automation use cases being built via Box Skills already.