Our workdays are getting noisier. Never-ending emails, text messages, constant notifications from more apps and more platforms—it’s disruptive and distracting. And then there’s content. All kinds of documents, spreadsheets, presentations, videos, and photos. Industry research shows that employees at larger organizations use an average of 36 cloud services at work, including tools for productivity, project management, communication, and storage. This information overload is a key source of pain for people at work—and a prime opportunity to leverage the help of machine intelligence.
How do we define machine intelligence?
When we talk about machine intelligence at Dropbox, we mean the whole range of applied machine learning, from simple linear classifiers to advanced deep learning networks. For many years we’ve been building a world-class machine learning team, improving our platform behind the scenes. We started with a lot of foundational work on image recognition to improve our users’ experience of organizing the massive amount of photos they keep on our platform. The fact that many of those photos have text in them led us to invest in our mobile document scanning capabilities and custom optical character recognition (OCR) pipeline to help our users quickly scan and find their content. We combined classical machine vision techniques with advanced deep learning methods to create a mobile photos experience that was faster and more accurate than any off-the-shelf solutions we could find.
A lot of the content in Dropbox is already in text-based documents, so search is another area of significant investment for us in terms of machine learning. We’ve now completely rebuilt our search infrastructure (details in a post coming soon) to improve the quality and speed of results from the hundreds of billions of pieces of content our users entrust to our platform. Because of our granular sharing permissions, each user has a unique corpus of documents to search within. This creates whole dimensions of personalization that web search engines largely ignore. Add to that activity signals related to relevance—such as files with recent comments or those recently viewed by team members—and you have a really productive use case for machine learning.
Each product innovation makes our users’ lives a little better, but how can we make larger leaps toward freeing people’s attention so they can find focus and flow at work? Instead of people working harder to keep up, can we make their tools smarter? To achieve this, we’re weaving our unique approach to machine intelligence into all of our products and surfaces. We call this effort the Dropbox intelligence initiative (DBXi), and we’re excited to share some updates.
What’s the big problem we’re trying to solve?
We see tremendous potential to use machine intelligence to improve the work experience itself. Quieting the noise for individuals is the first step toward helping people in organizations work better together. There are lots of engineering challenges behind creating an intelligent workspace, but the motivation comes from design and product insights. Our researchers have invested years studying how the rise of SaaS applications and mobile have changed the way contemporary knowledge workers do their jobs and run their teams.
This research shows three kinds of activity that people spend way too much time on:
- Organization: Work content is spread across files and cloud content in different silos. We spend a lot of time just finding what we need before we can start working on it.
- Contextualization: It’s hard to put content together with the communication about it. Constantly shifting context requires mental overhead and cuts into our ability to think clearly, leading to stress and burnout.
- Prioritization: With everything so scattered and fragmented, it’s hard to know what’s really important.
These activities are an increasing part of modern work, and they all get in the way of maintaining focus. And these problems are compounded for every team member as well, so there’s a lot to gain from taming this complexity. We think our intelligence solutions can help address these pain points at work.
What’s our next step?
One example of a user experience we’re exploring as part of the DBXi initiative is demonstrated in the animation above. In this prototype, we’ve evolved the existing desktop surface where our users check sync activity and see notifications into a more dynamic view that intelligently highlights their most important work connected to Dropbox.
We suggest the most relevant content by traversing a user-specific graph that connects people, content, and activity signals in privacy-preserving ways. This includes not only files but also content like Google Docs, as well as collaborative activity in emails, messaging apps, and calendars—whatever a user chooses to connect to Dropbox. And we make all this content searchable in this surface as well. To cut through the noise, we prioritize notifications and show content related to calendar events in personalized activity feeds. And to declutter these feeds, we cluster content and collaborative activity across silos so users can immediately see which projects need their attention and be only a click away from the content they need.
This prototype is only one of many surfaces we’re exploring for intelligence. The data graphs and models we build for products and search are generally reusable across surfaces and features. By scaling our internal machine intelligence platform, DBXi is multiplying the efforts of our dedicated intelligence product team so all our engineers can modify and validate models for intelligent features, improved search, and other business optimizations. These scalable methods and common infrastructure give our product managers and designers the flexibility they need to experiment and take chances on novel directions.
From a technical perspective, these are important problems to solve, and success means not only intuitive user interfaces, but also lightning-fast response times, industry-leading prediction, and the highest standards for maintaining data privacy. Want to help us build the next-generation intelligent workspace? We’re hiring!