Some tech CEOs run the show but don’t really understand the nuts and bolts of the company’s product. Not Juan Carlos Riveiro. With more than 100 patents, he’s clearly both a techie and a business leader.
Riveiro is the CEO of Vilynx, a firm which uses artificial intelligence (AI) and machine learning to help media companies make video “smarter.”
To date, the Spanish company has received about $15 million in funding from European and North American venture capital and angel investors. Vilynx has offices in Barcelona, Palo Alto and New York City.
This interview has been edited for brevity and clarity.
Former CBS and CNN executive Joe Klein joined Vylinx this past fall as President and said, “We are the only artificial intelligence company with a platform focused exclusively on media.” Can you substantiate that hefty claim?
Our main market differentiator is that we are the only ones with a true AI brain that not only recognizes but can actually understand. Our brain is self-learning and provides true intelligence versus the rest of the industry that provides mostly recognition.
There are plenty of companies that address one piece or another of AI for media companies, but they’re simply one-off products. Examples include Taboola for recommendations or Minute.ly for autogenerated previews.
Vilynx is a platform, and none of the companies currently working in the AI space for media offer clients a platform. While Google AI or IBM’s Watson provide their services to the media industry, they also work within many other verticals.
Additionally, Vilynx is the only company that has made it its mission to apply AI towards all of the specific-use cases of media—from indexing raw content at the point of entry, to building intelligence for content management and editing systems, and the user-facing side with content personalization, discovery and engagement.
We have media professionals on our team with experience working on all of these content touch points, who understand the challenges of creating and distributing content and the opportunities for improvement via AI.
How can AI become indispensable to media companies?
To be truly valuable to media companies, AI must offer an understanding of their content—not just reams of crunched data or recognition of faces and buildings. That’s exactly what the self-learning Vilynx brain does. Yes, it analyzes faces, processes speech, reads text and identifies buildings and places—but then it takes the crucial step of putting all this into context.
So, it knows that Justice Kavanaugh is a much more polarizing figure than Judge Kavanaugh was one month ago; it knows who’s streaking toward a World Series, not just who made the playoffs; it knows the difference between Tyrion Lannister and the actor who plays him, Peter Dinklage.
It then applies that insight to a range of products that engage viewers more deeply, including auto-generated previews, precision video recommendations, personalized user experiences, and targeted marketing messaging.
The self-learning Vilynx brain then accelerates the production process while reducing costs and freeing humans up for real value-adding contributions via extensive metatagging, database search, translation, and closed captioning.
How will machine learning prove valuable for the media industry?
The average video clip on a media website is tagged by hand with fewer than five fields of data. The Vilynx brain provides more than 50 fields, using facial recognition, natural language processing, object identification and text analysis.
It’s crucial for content creators to understand their audience. Right now, editorial teams get lists of video/article headlines that do well on a daily/weekly basis—nothing more. This is superficial and there’s not a whole lot of insight you can take away from this.
Machine learning, and, in our case, a dynamic knowledge graph, allow you to really understand what each piece of content is about with metadata. Combine that with engagement metrics to figure out what is really resonating with your audience—with clarity and detail. That is something that’s never been possible and available to content creators.
Talk about your company’s partnership with NBC and MSNBC.
The first step for these properties was to leverage Vilynx’s content recommendations across all video pages, and then across all articles. Editors had been manually adding related content for every story published. The fact that this is now automated gives considerable time back to the editors. We are helping them personalize every point of contact with users.
How will personalization in AI alter content consumption and monetization?
We are moving to a more surgically-precise world in which advertisers aren’t buying eyeballs en masse. They’re buying individuals and paying a premium for the attention of exactly who they want.
In the subscription world, Netflix uses personalization to make increasingly accurate recommendations. Compare that experience, which is powered by AI, to their rivals who use “AO”—alphabetical order—when listing their shows.
Publishers have thousands or millions of people interacting with their products. There’s no way that a static list of content manually curated or alphabetically ordered can effectively engage all your different users.
The common editorial approach is to guess what content will generally resonate with most of your audience, but the reality is that if you’re targeting everybody, you’re targeting nobody. I don’t think it always needs to be 100% automated either, depending on the content.
The best approach in some scenarios might very well be a combination of editorial judgement plus algorithms, a/k/a “algotorial.” News is a good example of this, because it is also meant to be informative.
There’s also a lot of potential for monetization. Targeting, in many cases, is rather rudimentary, based on a few socioeconomic or demographic tags, or maybe a specific section or series of your content.
But maybe the person you’re interested in reaching, as an advertiser, is watching something else, and you’re totally missing them. Or maybe you’re broadly targeting one demographic group, but there’s another that’s also interested in what you have to offer. These are shortcomings of the systems many publishers use today and can be addressed with deeper content/user understanding.
I met your CTO Elisenda Bou-Balust at the NAB New York show a couple of months ago. She too preached the importance of “self-learning brains.” Why are they vital for the media industry?
A supervised brain is only able to recognize the things it was trained for, so it doesn’t evolve. A supervised-learning brain and a self-learning brain like Vilynx’s can both recognize a new iPhone XS.
However, Vilynx’s self-learning brain also will know that it’s controversial because it’s so expensive, isn’t thought to be a significant leap forward in hardware, and that it has generated questions regarding whether Tim Cook is up to the task of filling Steve Jobs’ shoes. When it comes to content, context is everything.
Only self-learning provides context. Self-learning means the brain evolves with time, is able to adapt to new events, to get more knowledgeable and smarter without any human involvement.