TECHNOLOGY

Why did we acquire Curio AI? Since the invention of object storage in the late 1990s, unstructured data has mushroomed and now comprises over 80% of all stored data. While structured data is defined as searchable, unstructured data is not. At least up until now.   

It hit me when I was talking to the Boston Red Sox about migrating their vast video archives (stored on tape) to the cloud where content could be instantly retrieved. The problem wasn’t so much the migration process itself. Rather, it was that they still wouldn’t be able to find anything once they migrated it. There was almost no metadata – sometimes just some pencil marks on the box.   

I realized that Artificial Intelligence now has the capability to recognize faces, logos, objects, and locations. It can read the names and numbers on a player’s jersey. It can interpret crowd noises and convert an announcer’s voiceover into searchable text. It can detect the sound of a bat connecting with a ball. And much more. And automated metadata creation can be done quickly, easily, and inexpensively.   

I am now convinced that object storage without metadata is like a large library without a card catalog. Without the metadata, the data itself is nearly useless. And just as a card catalog is integral to a library, metadata should be part of the storage itself. Wasabi should not only store the raw data but also create the index that makes the raw data useful.   

So, I am very pleased to announce Wasabi’s acquisition of Curio AI from GrayMeta. We have combined our low-cost, high-performance object storage with Curio’s AI to create a new class of intelligent storage. You simply upload your video to Wasabi Curio storage, and Curio AI automatically creates second-by-second metadata. It can also index other media such as photos, audio, and PDFs. Several major league sporting organizations across baseball, futbol/soccer, basketball, and hockey are already seeing the benefits in early trials.  

I believe that AI-powered object storage may be the biggest advance in data storage since the invention of object storage itself. It’s clear to me that within a few years, all object storage will incorporate AI-powered indexing. I am hopeful that Curio’s ability to generate detailed metadata will finally incentivize the media and entertainment industry to migrate those hundreds of exabytes of archives to the cloud.