Google's new Photos app doesn't just store and organize your photos — it actually knows who and what are in them.
It knows what your dog looks like, when and where you last ate pizza, and can automatically pull all of the photos of your 9-year-old niece from the time she was first born up until her most recent birthday into a neat album.
Google is hoping this new Photos app becomes the single digital hub for all of your memories, and it seeks to differentiate itself from rivals like Apple and Dropbox in two ways: by making your photos just as searchable as an email in Gmail or a link on the web, and by offering free unlimited storage for photos that are 16-megapixels in size and under.
"We kind of decided about a year ago that we wanted to rebuild things from the ground up," David Lieb, the product lead for Google Photos, told Business Insider.
An app that gets smarter the more you use it
Google uses machine learning to power the scary-accurate image recognition within the Photos app, and based on the demo I've seen, it's incredibly fast and accurate. Other services let you search photos based on when they were taken, where they were taken, and how you've tagged them.
But with Photos, you can simply type in the word "pizza," and any photos you've taken of pizza will pop up whether you've labeled the pictures or not. If you went rock climbing last summer, you can type in the phrase "rock climbing" to find those images. Google also automatically sorts photos based on these types of phrases in the Categories section. So, all of the photos of any given person will appear in one organized album, all photos of food would be in another, etc.
Google already uses machine learning — a term that refers to a type of algorithm that learns on its own without human intervention — for many of its products, including Google Now, Google Maps, and Search. But now we're seeing it being applied to photo storage.
There were two aspects that Google focused on the most when developing the new Photos app: making sure you can search using broad words and phrases that feel natural rather than keywords, and making sure the app is easy to use.
"The design of the app is probably the thing that spent the most time iterating and getting right," Lieb said.
To decide which features should be included in the app, Google brought in outside testers to determine what would resonate with actual users — a common tactic in most product development. Google also spent time testing paper prototypes of the app for months to get the design right.
"We put [the app] in front of real users and said please take these components we put in front of you and construct the app you want," Lieb said.
That also means dozens of features were cut from the app that didn't make it into the final version. Google initially tested a layout that was similar to the Gallery app on Android phones — it would show you the folders on your phone like, Camera, Downloads, etc. But testers didn't seem to be into that layout — 98% of them would just click the Camera category and ignore the other ones, Lieb said.
"So we decided to show all of those photos in a single view so the user doesn't have to decide," he said.
But, layout aside, it's really Google's advanced machine learning technology that powers the app.
"At Google I/O, you heard Sundar (Google's SVP of product) talk on stage about how Google has been making machine learning investments for decades now," Lieb said. "So we do actually have the best capability in the machine learning realm. Applying that to photos is one example, but we have many."
The new Photos app is available for download on the desktop, iOS, and Android.
SEE ALSO: Google just took the lead in the dangerous game called 'Race To Zero'
Join the conversation about this story »
NOW WATCH: This new documentary reveals how taking photos of everything defines our lives now