The internet acts as a vast archive for our memories—a massive trove of information that we no longer have to store ourselves, because Google is often faster than our own recall.
But while the internet has been a boon for stowing information, in other ways the digital world remains harder than ever to navigate. When countless websites and applications and files are spread across multiple devices, it can be difficult to remember where, for example, you came across a specific article. Did you save the link in a Google Doc? Did you bookmark it? Was it on Facebook? Did someone text it to you?
This is the problem that the startup Atlas Informatics is trying to solve. The company, founded and run by the former Napster cofounder Jordan Ritter in 2014 as part of his Ivy Softworks Innovation Studio—is launching its first product, a universal search application called Recall. Recall was designed with the strengths and weaknesses of human recollection in mind, acting as a kind of photographic imprint of everything you access across your devices.
Based on the company's research, about 20% of our queries are for things we've already seen. The search engine works by capturing and indexing anything you see on your computer or smartphone screen. It takes screenshots of every window, every file, every text, every Tweet, and every Facebook post you look at, and then allows you to search for it (and access it, if it's a file or web page). Recall also helps improve other search engines, adding in a sidebar next to your Google or Spotlight search results that shows your Recall results.
Fascinatingly, Recall's functionality is based on our computers' current accessibility features. According to Ritter, Atlas can search across all platforms thanks to the same technology that allows your computer to read the words on a screen aloud, or translate them into braille. So instead of integrating with the APIs of every platform on a user's computer, Recall works by integrating with the device itself in order to record every single thing you look at (unless you hit pause).
Privacy is a natural concern when it comes to such an expansive application. The company has a clear policy on how it approaches this issue—mainly that you can delete any single piece of data or all of your data at any time, that it is encrypted both at rest and in motion. A simple delete button lets you remove any record of an item from Atlas's system and adds it to a blocked list, which you can also add to proactively. The system also has a pause button in case you don't want it to see, for instance, your bank accounts.
While this is smart technology, much of Recall's appeal comes from the design of its user interface. One key element is that it's a visual search app. Because we can recognize things we've seen already very quickly, Recall shows the user thumbnail images of the things they could be looking for. These thumbnails are arranged in an organic, amoeba-like shape, with clusters of images grouped at the edges of the main search results that are usually sorted by item type, allowing the user to click into them to narrow down the search. The more relevant the system thinks the search result is, the larger it appears. "As long as it’s your own data, it resonates with you really strongly because you recognize it," Ritter says.
It took the company's designers a long time to land on the amoeba and its clusters. Initially, they visualized search results as a timeline, segmented by format type. When users found that difficult to use, they introduced thumbnails to it—but even so, the data design felt like something from a science textbook.
Instead, led by VP of design Matthew Holloway, they began to explore other forms of information design. They turned to science fiction films and television shows like The Matrix, Minority Report, and Star Trek, and looked at the way information was laid out in museums like San Francisco's Exploratorium and Austria's Ars Electronica. Delving into visualizations from the health care industry and finance, along with transit and urban mapping, provided even more precedent. Eventually they landed on a more organic form, organized around galaxies and constellations that you can enter and explore.
The clusters were inspired by the associative nature of memory: You might not remember the exact location of the article you're looking for, but perhaps you remember the song that was playing in the background when you saw it, or that this picture of your friend was on the screen at the same time, or that you were sitting in Starbucks. Search for that song, and Recall will pull up other documents that were on the screen at the same time.
"We build a latticework of memories based on concurrence, location, people, large pillars in our lives," Holloway says. "Human memory can be quite infinite, and we didn’t want to prescribe a structure for how you should remember things."
Especially because we often remember things wrong. "We can oftentimes mis-associate things with each other," Holloway says. "So we wanted to build the user interface to take advantage of those different associations so you don’t end up beating the same path and you can still get to where you want to go."
Recall's deep integration with your computer is meant to feel like more than just another application, which is why the app is designed to be an immersive, full-screen experience. It's supposed to revolutionize the way you think about your data. "When the save button is obsoleted, it’s profound the shift in the way you deal with your devices and data," says Ritter. "When we fully deliver on the value proposition, you’ll lose that version of battery anxiety."
Atlas Recall is now in open beta and can be downloaded for free on Mac and Windows. The company plans to implement a freemium model once Recall comes out of beta, but with limitations on how far back its memory will reach. More functionality, like the ability to search by location, is coming soon. Ritter imagines that next year's version of Recall will start offering smart suggestions, using context about what you're doing to anticipate your needs.
"We’ll start with personal search, but what we’re actually trying to do is make this thing know us. It’s still stupid. It sees everything you see, it sees everything you do, it has context. It just doesn’t use it and put the pieces together," Ritter says. "That’s what we’re trying to do. How could this thing become an extension of me instead of a tool I have to work?"