“Viewers are spending 25% or more of screen-time looking for something to watch. Solving this issue is our major focus at Vionlabs and early customer engagement has shown that our AI-powered content analysis approach delivers game-changing results,” said Marcus Bergström, CEO, Vionlabs.
At deployment, Vionlabs measures everything in the videos in an operator’s catalogue and trains its AI engines to work out what matters in each asset. The platform uses AI and deep learning and combines this video analysis with a viewer’s detailed watch-history to provide content discovery services to operators through a cloud-based SaaS model. The platform is designed for linear TV, catch-up, streaming and all flavours of VOD.
“Our solution is based on deep thinking from our engineers and their key realisation that we could train AI engines to learn what variables should be measured in the video and audio and combine this with viewers’ watch history to significantly out-perform existing content discovery solutions. Our platform is now providing live content discovery that improves critical operator metrics such as VOD buy-rates and customer engagement and we’re excited to make our public debut at IBC2019.”
The Vionlabs content discovery platform analyses each video in detail and combine this with the viewer’s watch-history. The platform doesn’t need to know anything more about the consumer and it doesn’t try to guess who the viewer is. All that matters is the content they watch, for how long and how often.
“During the year we have productised our technology and deployed our content discovery platform for operators that are now enjoying a significant uplift in VOD buy-rates and engagement,” said Giles Wilson, CTO, Vionlabs. “By serving up the right content at the right time, our content discovery platform maximises the time viewers spend watching the content they love and minimises the time they spend searching for it. Leading to happier viewers that spend more time and money on a video platform.”
Vionlabs has trained multiple AI engines to measure everything in the video, including object recognition using computer vision as well as colours, pace, audio, and many more variables to produce a fingerprint timeline throughout the content.
The AI engine learns what matters and how changes in these fingerprint timelines are connected to what content individual viewers enjoy. Knowing similarities between content is vital, but it’s not enough on its own. Vionlabs has also trained an AI engine to analyse a viewer’s watch-history. Finally, it has an AI engine that takes the outputs of the other AI engines to provide the most accurate content discovery platform.
The content discovery results produced by the Vionlabs platform are made available through a cloud-based SaaS model integrated with the operators back office systems. The Vionlabs platform is already deployed and is generating very significant uplifts in VOD buy-rates and viewer engagement for its operator customers.