Hitmetrix - User behavior analytics & recording

With Syte, AI Lets You Shop Almost Anything You See

With the responsiveness to indications of taste set by Amazon, many sites now offer browsers suggestions for additional items that resemble the style of the item initially selected. Now with the advance of AI and object recognition, it is possible to get that kind of recommendation from an image anywhere, whether it’s on a retailer site, a style article, or even something captured on your phone from real life.

Partnering with technology companies like SAP, Naver, Microsoft, Line, and Oracle, Syte.ai pulls together object recognition, AI, and machine learning to render anything visual “clickable and shoppable.” If you can get a picture of it, you can shop for it.

Syte.ai is designed to grasp all elements in a still or video image and render them shoppable without having to work thought text and tags. The advantage for those shopping is that they just have to locate or snap a picture of the item they want and let Syte find the item or something very close to it for sale online.

For retailers and brands, it opens up new ways to reach customers who are interested in the styles they carry even if they identified the particular style on another site. If an item on their own site is out of stock or not available in the size and color requested — a situation that Syte says accounts for $634.1 billion in lost retail sales — its style matching can surface similar things to offer the customers.

For publishers, Syte opens up a new stream of revenue from shoppable images. Syte pays publishers based on a revenue share model.  It also allows for publishers to feature sponsored posts using visual search, displaying results for targeted brands.

Syte’s VisualBot can be hosted on a brand’s app, website or Facebook page. 

Here’s a video of Syte’s effect for fashion:


The company also provided me with pictures that show what is being searched and what results Syte offers as you can see below.  In this particular picture, no shoes are visible, but where they are, there would be a box for them, as well, in case that is the item visible on the image that you wished to shop for.

I wanted to test this out for myself, so I downloaded the Syte Inspire Chrome Extension (Android users can download the mobile app) and started shopping online to see how it works. The results were better for some things than others. I went on Saks Fifth Avenue’s site and selected this skirt.

I clicked on the Syte extension, and returned two choices that were close. You do see four skirts in the screenshot I took, but two are of the original one.

However, sometimes it takes more than a single click. When I tried the same with a pair of shoes, it didn’t register them automatically, taking a read of everything from the page equally. I then had to upload the picture I had downloaded from the site to get it to fetch similar styles. It did return quite a number of those, though none could be mistaken for the original because of the distinctive detail on it.

I played around some more and discovered that some sites don’t function as well with Syte. The Bloomingdale’s site does detect some action because when I try to put in Syte: It offers connect me with a sales associate. But for the four different items I tried it on, the Syte operation would not go through, even though it was picking up the image from the page. I could only get results when I saved the Bloomingdale’s image to my computer and then uploaded it on Syte to fetch matches.

It’s possible that Syte is still working on this to make the user experience more consistent across sites. But it’s also possible that some retailers don’t want their sites to be so easily shoppable beyond their own stock.

Related Posts