Box on the future of enterprise content, workflow and intelligent search

Cloud content management provider Box has been acquiring talent and technology in recent months to build up its capabilities in workflow and intelligent search. I spoke to Chief Product Officer Jeetu Patel on a recent visit to London about the company's plans in these areas as well as the future direction of content in the enterprise.

I've written in the past about the role of content in the context of digital collaboration across the enterprise. For collaboration to be purposeful, content has to be tied into other functions and applications, particularly workflow, search and messaging. Therefore, by acquiring new skills in those areas, and by partnering on messaging with Slack, Facebook, Google, Microsoft and Cisco, Box is ensuring that its content platform delivers the goods in the connected enterprise.

Defining enterprise content

The definition of content is also evolving. When we think of enterprise content, we often picture it in terms of traditional documents and files, but Box is seeing a lot of growth in image, audio and video files. Customers often focus on these types of content because Box has various AI tools that can automatically extract useful data and metadata, explains Patel:

You have to have an intersection of three things. Where is there a lot of data? What are customers asking for problems to get solved, and you could have a good return? Are there algorithms in place that can do a good job at going out and assessing and analyzing them? On all three of them, what we found was image, video and audio came out at the top, which is why we said let's start with those.

There are use cases such as extracting and validating the data from images of driver's licenses, categorizing images in a library of marketing files, or making transcripts of call center conversations and online meetings. Box has partnered with IBM, Google and Amazon to use their AI models for this analysis. "The beauty of this is that if one AI algorithm gets better than the others, you just switch it," Patel notes. The aim is to maximize the value of content when it's stored in Box:

The goal is not just to have content for the sake of having content. It's so we can constantly compound the value for content. Every single time that you go out and apply a model to content, there's value that's been being compounded.

There's a constant level of compounding that we want to do so that as the content ages, the value of the content goes up in the system — rather than today, once the content goes past 120 days, you see the value actually go down because no one uses it. Archive content should have the most amount of value in the future, because you have more models applied to it.

Intelligent search

Once the content has been tagged, the next challenge is making it discoverable so that people find it when it's going to be of value. That's where Box's most recent acquisition comes in. Last month, Box acquired the team from Butter.ai, who have been using machine learning to help surface relevant content in the enterprise. Their skills will help searching in Box become more contextualized, predictive, and personalized, as Patel explains in a blog post:

For example, searching for the word 'contract' in Box for a member of the sales team will be more likely to turn up sales contracts whereas the HR team will be more likely to see employment contracts; or a retailer searching for 'blue dress' would return images of blue dresses from the latest clothing line.

This will complement the work Box has already been doing to build a graph that maps the relationships between people and content, which helps determine what should be surfaced in each user's content feed. This is important to help people organize their work, says Patel:

As your corpus of content gets bigger, you want to make sure that you're not just going out and browsing through content because it's very hard to find, or searching through content because you might not know what you're searching for. If work is getting done that you should be knowing about, then that should surface up to you — if people have chosen to share that with you.

Workflow - Progressly and Relay

The other recent acquisition added new workflow capabilities. Progressly, which described itself as "the world’s first cloud-based Operational Performance Management solution," was acquired in May, with its Co-founder and CEO Nick Candito being named Box's General Manager of Workflow. Progressly secured $6 million Series A funding in 2016 and was founded in 2014. Candito along with CTO Ruslan Belkin had both previously been part of the team at sales intelligence startup RelateIQ when it was acquired by Salesforce. Ruslan subsequently went on to become VP engineering at Metamind, which Salesforce also acquired as part of its development of its Einstein AI toolset.

Box already has workflow capabilities as a result of its partnership with IBM. This led to the creation of Box Relay, which is already in production with customers. Box Relay added three new features last week — pre-built templates, an API, and the ability to automatically trigger a workflow process when a user or external system moves or uploads a new piece of content into a Box folder.

Types of workflow automation

So why acquire more workflow capabilities when Box Relay is already in place? It appears that Box wants to increase the native workflow automation that's available within Box, while continuing to expand Box Relay's ability to connect out to external systems. Patel explains that Box needs to cater for several different types of workflow automation:

I think there's different segments of the workflow market. One segment of the workflow market says I am just doing some ad hoc automation.

Then there's a process that I need to go out and implement and then you have to have a set of deterministic steps that you have to go through.

Then there's, I'm going to go out and integrate with a bunch of different systems and make sure that you can trigger things from one system to the other. So when an event hits Box, go ahead and trigger something in Salesforce and then if something happens in Salesforce, go ahead and trigger something in Workday.

Then you have hardcore BPM engines that can actually go out and fundamentally re-architect a business process — companies like Pega, Nintex and all of those.

We feel like we need to have a solution in each one of those categories. Some through partnering, some by organic, some through making sure that we integrate with third parties.

In the enterprise market, connections into legacy systems are also important, particularly when an organization wants to move content into the cloud while some of its processes remain in existing applications. Patel elaborates:

Where content needs to be in the cloud, we want to still make sure that that business process doesn't break as a result of it. Our customers have said, 'Look, just because you're moving to the cloud doesn't mean that everything else will move to the cloud.

We're starting with content, but we might actually keep some of the other pieces on-prem right now. As we are keeping those on-prem, we still want you to work with them.

Workflow seemed like a very easy one to work with because the technologies that we have make it very easy for us to trigger an event based on an action and then make sure there's a web hook into a different system.

There may be other acquisitions waiting in the wings — in an interview at the end of May, Box CEO Aaron Levie spoke of the likelihood of small acquisitions as part of its investments in the fields of artificial intelligence, workflow and security.

My take

The acquisition of Progressly seems to be quite a significant move in terms of extending Box's workflow automation capabilities. Evidently the partnership with IBM is not delivering all of what Box feels it wants to do, although as Patel points out, there are a lot of segments to the workflow market, and it may be unreasonable to expect one intiative to cover them all. No doubt we'll find out more at the company's annual BoxWorks conference at the end of this month.

Both these acquisitions bring AI expertise into the company and there again it appears that Box wants to build up its native capabilities in addition to what it's getting out of partnerships with AI providers.

Discovery and workflow are both highly complementary to Box's core content platform, adding extra substance to the company's proposition, so the acquisitions make a lot of sense. They signal that Box is preserving its differentiation as the pureplay cloud content platform most attuned to enterprise priorities.