The Media Intelligence Stack: A Blueprint for the Future of Content Operations

As streaming dominates TV, media & entertainment companies are rethinking the viewer experience to boost engagement and advertising revenue. Disney’s ESPN is imagining a more personalized, AI-driven SportsCenter, allowing fans to tailor their sports viewing experiences and enhance engagement through customized content delivery. By leveraging AI for real-time data analysis, SportsCenter could dynamically adjust highlights, commentary, and recommended clips based on a viewer’s preferences and past interactions. This approach moves beyond simple personalization into creating brand new experiences designed for the individual viewer.

Similarly, David Zaslav, CEO of Warner Bros. Discovery, has emphasized the company’s commitment to leverage artificial intelligence to enhance content discovery and viewer engagement: “We’ve been using AI and machine learning to personalize content discovery,” Zaslav said. “We continuously innovate to improve our models to present the right content in front of our consumers at the right time. And this is helping us to drive better content diversity on Max.”

Personalized, shareable, and unique experiences are the future, but for most media companies, this future relies on outdated technology. Legacy media systems were designed for linear and static content distribution, making them ill-equipped for the new 1-to-1 experiences required in the on-demand era. When we talk to technology leaders at media companies today, everyone is using AI in media intelligence, but there are significant challenges:

• Video content can quickly reach petabytes of data. Storage and processing are expensive, and every tech leader is trying to lower those costs.
• Preparing content for processing by the foundation models is an expensive process requiring sampling, file formatting, normalization, and audio alignment.
• Media companies are either using machine learning pipelines to extract features from visual content or exploring foundation models. The landscape is evolving quickly and everyone is thinking about how to stay relevant.
• Once foundation models process content, vectors and embeddings need to be stored in a structured format, typically within a database or a specialized storage system designed for high-dimensional data. This allows for efficient retrieval and processing of the embeddings for various tasks such as similarity search or machine learning applications.
• How can teams improve the data over time? Powering recommendation engines requires massive amounts of accurate metadata labels. Ideally, it’s not solely the responsibility of the machine learning team to manage that process.

‍

Most companies’ media intelligence stack is loosely held together by in-house technologies, legacy machine learning pipelines, and some AI. Different business units within the organization often solve the same problems with overlapping efforts. Even in the same business unit, different departments duplicate efforts, whether processing visual content multiple times or repeated tagging efforts for various campaigns.

There is a better way.

The Multimodal AI Platform and the future of content operations

At Coactive, we believe the best approach to the media intelligence stack is to unify on a single platform, The Multimodal AI Platform. As the industry shifts toward hyper-personalized, 1-to-1 experiences Disney and Warner Brothers are discussing, the use cases for visual content are expanding. The best way to support this is to automate and streamline the operations that power visual content services.

The Multimodal AI Platform is a complete platform that meets companies at any stage of their journey. It enables efficient processing, analysis, and use of visual content. Instead of treating personalization, search, monetization, and compliance as separate efforts, a single intelligence layer supports this variety of use cases in a streamlined and scalable manner.

‍

‍

Let’s examine these use cases in detail.

AI-driven content personalization

As we already covered, audiences increasingly expect content tailored to their preferences, behaviors, and viewing habits. Traditional recommendation methods, based on broad demographic assumptions, are no longer enough to keep viewers engaged. Netflix executives emphasize that their recommendation engine drives user retention, with over 80% of hours watched from personalized recommendations.

Major platforms like TikTok, Instagram, and Spotify rely on structured content intelligence to surface relevant media to individual users. These experiences require extensive customization of metadata. The manual tagging and fragmented archives that worked for the early era of personalization now limit media companies’ ability to compete with algorithm-driven platforms.

The Solution: We see the Multimodal AI platform as a way for media companies to close the gap between traditional and streaming-native platforms. Instead of competing for expensive technical talent, they can adopt a single platform and focus existing resources on building recommendation engines and personalized experiences on that foundation.

Search and content discovery that understands user intent

Media organizations spend millions on content but struggle to find and reuse their assets effectively. The problem isn’t just tagging. Historically, machines can’t see and understand content like a human editor.

The BBC is using AI-driven archival search to enhance its ability to retrieve historical footage on demand. By applying machine learning models to index video content, the organization has significantly reduced the time to surface relevant clips for news segments and documentaries. This shift has improved newsroom efficiency and enabled richer storytelling by integrating archival content into modern reporting. Every new advance in AI increases the pressure on organizations to deliver improved search and discovery.

The Solution: Advances in multimodal AI foundation models are happening at a remarkable pace, outpacing machine learning. To accelerate transformation, media companies should process all visual content with multimodal AI and make it available via the data layer of the Multimodal AI Platform for various use cases including search and discovery.

Monetization and ad targeting

In early 2024, NBCUniversal announced its “One Platform Total Measurement” framework. “This new capability analyzes massive amounts of content across the NBCUniversal portfolio paired with the company’s extensive first-party data sets to produce emotion-based, AI-powered audience segments.” As traditional ad models erode, media companies must extract precise contextual metadata for advanced advertising. Advertisers want to target moments, not just shows, placing ads against specific scenes, emotions, or themes that align with their brand.

Solution: AI-driven contextual targeting will be crucial for media monetization, aligning premium ad inventory with advertiser intent, improving CPMs and engagement metrics. Leading media organizations are exploring AI-based metadata enrichment integration into their ad systems to grow their ad inventory and deliver improved outcomes for advertisers.

Brand trust and safety

For content-driven businesses, it’s critical to provide advertisers confidence that their brand will appear in complementary contexts. Fandom is the world’s largest community platform with over 300 million user uploads monthly. While most are welcome additions to the community, a small percentage come from bad actors sharing content that violates the community guidelines. This type of content erodes the community and damages brand trust.

Solution: Using Coactive’s Multimodal AI Platform, Fandom has automated content moderation, reducing the time to remove offensive imagery from 24 hours to seconds. Fandom’s VP of Engineering & Data Science, Florent Blachot, views the investment in this unified intelligence layer as an accelerator for visual content workloads. The team can use this workflow to prompt users with feedback on uploads, like flagging a low-quality image.

The future of content operations

Media technology leaders recognize a critical truth: all media intelligence use cases—personalization, ad targeting, search, and brand trust & safety—share the same mechanics. Instead of building separate solutions for each problem, forward-thinking media companies want a single platform that integrates everything and powers various end use cases.

By standardizing on a single platform, organizations avoid redundant work and provide developers with a scalable foundation for building new applications. This approach accelerates innovation, reduces operational costs, and ensures that any AI-driven workflow—whether for content discovery, compliance, or monetization—builds on the same core intelligence infrastructure.

For media companies, the question is no longer if they need AI-powered intelligence, but how to implement it to scale across their content operation. Viewers expect highly personalized, 1-to-1 experiences and shorter, shareable content. Delivering on this next phase of the personalization journey requires a technology foundation for content discovery, accurate metadata, and petabyte scale across massive content libraries.

‍