Enhancing Content Discovery Through AI Collaboration
Today, viewers are inundated with an ever-expanding sea of content. The challenge for both consumers and service providers is no longer just about having access to a vast library of shows and movies, but rather about efficiently navigating this to find truly relevant and engaging material. This is where the partnership between XroadMedia and Media Distillery comes into play, making sure that users find the most relevant content and quicker.
We caught up with the CEO, Roland Sars, from Media Distillery to chat about all things AI, metadata and how we have worked together to test and prove how our combined services can enhance viewing experiences for NLZIET’s subscribers.
Read the interview below.
A brief introduction to yourself and Media Distillery, including the kinds of projects you undertake. Do you have any exciting product updates?
Hi, I’m Roland Sars, CEO of Media Distillery. We’re video analysis specialists and in 2024 we’ve been celebrating our 10th year of using AI and Machine Learning to distill insights from live video streams.
Our solution suite is now quite extensive, but all our solutions share one common goal – helping to reduce the “time to content” for video viewers. Sometimes that means finding the precise start and end of a programme in a broadcast stream so that replay and catchup viewers don’t have to waste valuable time scrolling through the end of the previous show before they can start watching their chosen programme. That’s done with our solution EPG Correction Distillery. The same detection of accurate timing is applied in our Ad Break Distillery solution which helps operators to pinpoint exactly where the ads fall in a replay of live content. They can use this data to enable paid ad-skipping for a more VOD-like experience for the user (and additional revenue for the operator!) or to replace the broadcast ad-breaks with something that’s more relevant or timely when you’re watching a replay.
The rest of our solutions dramatically improve the content discovery user experience with the provision of enhanced metadata and media using Generative AI. Topic Distillery automatically identifies different topics, themes, people or locations that are mentioned in a broadcast stream and pinpoints the moments where they’re mentioned. This data can be used to create “chapters” that are displayed to the customer in the UI, so they can jump straight to the chapter that interests them most – like the specific feature about Taylor Swift in a chat show or the highlights of their football team’s match in a long programme.
Image Distillery generates interesting stills of a show from the video stream which can be used to illustrate the content in a replay UI and give viewers a visual clue of what the specific episode is about. And we go one step further with our latest launch, Preview Distillery which once again leverages AI and machine learning to create short, tailored video clips in near real-time from a broadcast stream. These previews seamlessly integrate into a platform’s content library, giving viewers an instant insight into every piece of content so they can quickly decide which shows are most likely to be to their tastes.
How do Media Distillery and XroadMedia currently work together? Why do our solutions work well together?
Together we have worked on a Proof of Concept (PoC) project for our joint customer NLZIET, an OTT streaming service in the Netherlands that highlights the use of both solutions, showing the added value of discovering hidden gems within their catalog. It utilizes our Topic Distillery solution in conjunction with XroadMedia’s personalization solution. Together, we can help viewers find content about topics or people they care about that would not have been available and discoverable otherwise. With user experience at the center of our efforts, our team is still working closely with NLZIET to apply the findings for the best product and further development.
How does the offering solve problems for service providers? What are the benefits? Any examples of projects?
Media Distillery started working alongside NLZIET in 2022 after they had already been working with XroadMedia. NLZIET was looking to explore the potential of AI in the video industry. This is what the PoC was set out to do. With Media Distillery, this collaboration tackles the lack of descriptive metadata – a notorious industry pain point, particularly in non-scripted live content like news, talk shows or reality shows. We focused on making content chapters visible in the play bar. This feature helps users to quickly see what the program is about and watch only the parts of the content they are interested in.
Topic Distillery automatically extracts metadata out of video, audio, teletext and EPG, utilizing Computer Vision, Speech-to-Text, and other AI/ML technologies. The real-time processing identifies topics and splits the content into logical segments, adding, on average, 21 topic tags per program.
With the successful implementation, TV operators can prevent churn among users with a preference for ‘snackable’ content, popularized by TikTok and YouTube Shorts.
“With Topic Distillery, this content suddenly, just 15 minutes after live broadcast, has long-lasting value – it’s a new way of consuming TV. We did consider other providers and technology solutions, but no one we spoke to could do what Topic Distillery does.”
NLZIET CTO, Sander Kouwenhoven
Furthermore, working with XroadMedia’s real-time solution helps the operator to present such relevant content to the right viewers instantly.
“XroadMedia has proven to be a great fit for NLZIET. The integration of the XroadMedia recommendation services was quick and seamless. The flexibility of their system was our main reason for choosing XroadMedia and that turned out to be a smart choice. We were able to quickly configure our recommendations to fit with our sometimes-complex business needs.”
NLZIET CTO, Sander Kouwenhoven
With the current hype around AI, have you seen an increase in demand for your solution?
Yes, demand has definitely increased along with the growing awareness of what can be done with AI. Companies are increasingly open to the possibilities and also the importance of ensuring scale and accuracy while also controlling costs.
How are you using AI in your service today?
We don’t just use AI and Machine Learning in our solutions. These technologies are the foundation of what we do. We set out 10 years ago with the goal of applying the latest advancements in speech recognition, image recognition and AI technology to enable video search capabilities similar to what Google had done for the web. By leveraging these technologies, we wanted to revolutionize the process of discovering and searching within video content. And that’s what we’ve been doing for 10 years – since long before anyone had heard of Chat GPT!
As the use of AI increases throughout the whole value chain for M&E, where does Media Distillery fit into it?
We have a bit of a mantra here at Media Distillery: “People are not looking to buy AI. They are looking for solutions to problems.” We don’t sell off-the-shelf AI tools that require heavy customization by a broadcaster or Pay TV operator’s own in-house team. We offer tailored solutions that directly address real industry needs, with AI as the enabler for improving content discovery and viewer retention.
You only have to visit a trade show like NAB or IBC to see that almost every vendor is now talking about AI, but much of it is limited to auto-transcription and auto-subtitling. While these capabilities have their place, they don’t fully address the needs of our customers. If you’re a Pay-TV or OTT operator that’s laser focused on customer retention, you need to keep your customers engaged with a high-quality UX. That’s the way to directly impact Net Promoter Score (NPS) and churn.
At Media Distillery, we focus on enhancing every step of the viewer’s journey within a video platform. Whether it’s generating tailored previews, ensuring content starts at the right moment, or helping users jump to the topic that interests them most in a longer show, we’re helping operators create seamless experiences that keep viewers engaged and satisfied.
Another differentiator for Media Distillery is that we aren’t using AI as a way to reduce headcount. Instead, we automate processes and solve problems that simply wasn’t possible to address before. Accurately identifying the start and end time of every advertising break in every TV show on every channel would be impossibly time-consuming for humans to achieve at scale, which is why they’ve not been done until now. In fact, I’d estimate that around 90% of what we do is not replacing existing technology or people; it’s creating new opportunities to enhance the viewer experience.
“People are not looking to buy AI. They are looking for solutions to problems”
Do you find that your customers are skeptical about the use of AI and concerned over its trustworthiness?
No. We typically do extensive proof-of-concept testing with customers, which helps them to see just how reliable our solutions are. This, combined with the success our current customers are having with the solutions, is a powerful demonstration that our AI solutions – which have been carefully tailored and trained for Media and Entertainment use cases – can be trusted. We’re delighted that most of our current customers have become ambassadors for our solutions which is also a great proof point for us.
Personalization done wrong can sometimes give users reasons not to trust a service, as recommendations don’t align with their interests and preferences. This is the same for some uses of AI.
XroadMedia’s AI solution provides personalization that can be automated or can be set to your business rules, keeping the control within your editorial teams. This way, the brand can still shine through but be able to tailor personalized experiences to viewers.
When it comes to personalization and trust, transparency is key. Service providers need to be clear about what data is used and how it is used.
Why XroadMedia? What is working well?
We are two companies that share a joint goal and a common commitment to improving user experience through better use of metadata. Making content discovery faster and less frustrating is essential to the success of any video service and the metadata that we generate with our video analysis has a big impact on that. But you also need to get that metadata to the UI in a timely way and without increasing the integration burden on the OTT, broadcast or Pay-TV service provider. By working with XroadMedia, we can streamline the delivery of metadata to our customers and onwards to their customers. Which in turn helps enhance the personalized experiences for end-users, improve user engagement with content discovery and match more content with relevant users.
“Making content discovery faster and less frustrating is essential to the success of any video service”
Which industry trends should be a focus for service providers this year? What challenges do they produce?
All the service providers we speak to are looking at how to maximize their margins in an era of ever-increasing competition. So, on the one hand, they’re looking at customer retention, and on the other, they’re looking at where they can control costs. We need to offer a value for money solution, which can be a challenge when you’re using AI at scale. But from the origins of Media Distillery, we’ve always had a focus on building for scale so that you can increase the volume of channels analyzed and data produced, without exploding the amount of resources required. I think that’s where we stand apart from some other parties that are trying to adopt AI and build use cases. When they actually go into production and they’re working with real amounts of content to cover a full platform, they find that the cost of cloud computing power suddenly comes as a massive shock. So I think you’ll see a lot of companies who aren’t so experienced with AI starting to back away from these solutions because they’re unable to use them in a cost-effective way.
What are the next steps in our partnership?
Personalization is becoming increasingly important for content discovery as video libraries grow and video service providers aim to avoid browsing fatigue. Effective personalization becomes much easier when you have detailed metadata about the topics, people, locations and events mentioned within the content, and especially when you can tie that information to specific timecodes. We’re excited to work together with XroadMedia on more customer projects and to keep elevating the discovery journey for customers in the media and entertainment sector.
Find out more about Media Distillery here, or get in touch with us to learn how we deliver personalized experiences for NLZIET.
