Unlocking Hidden Value: How AI Transforms Media Archives into Revenue Engines

• AI is revolutionizing how media organizations unlock value from their archives by automating the indexing, search, and retrieval of video content. 

• Traditional metadata systems fall short in surfacing specific moments, especially in older footage, but AI-powered solutions — like those developed by FPT — can analyze and tag content with precision, enabling natural language queries and dynamic clip generation. 

• By partnering with Akamai, FPT uses distributed infrastructure to optimize performance, reduce latency, and scale efficiently. 

• This transformation turns static archives into monetizable assets, enhances user engagement, and opens new revenue streams, while paving the way for agentic AI systems that create personalized workflows and content experiences.

The media and entertainment industry sits on a trove of archived content, yet many organizations struggle to unlock its value. Traditional search relies on basic metadata, like actors and directors for films or teams and matchups for sports, which makes finding exact moments when or where specific actions occur difficult. 

FPT, with their expertise in media supply chain, artificial intelligence (AI), cloud technologies, and digital transformation, can help media organizations overcome these challenges by using AI-driven solutions to enhance their content indexing and search capabilities, more efficiently analyze and identify key moments in vast archives, and ultimately unlock new revenue streams from their existing assets.

Consider the limitations of current video archive and content management systems: Although newer content may have richer metadata, it's still not detailed enough for precise retrieval. 

Older archives, which can span decades, are even more challenging. These archives often lack comprehensive tagging for teams, players, scores, or notable actions at specific time stamps, especially in sports

While manual effort can potentially address the issue, the process required is onerous, labor intensive, and uneconomical at scale. Simply put, indexing decades of footage by hand is not a viable or cost-effective option. 

With the proliferation and advancement of AI, a new solution has emerged that is transforming how organizations interact with media archives. Modern AI-powered systems can be built to automatically analyze, index, and improve the searchability of vast libraries of video content with unprecedented precision.

There are three steps in these modern AI-powered systems:

1. Model training and development

2. Content ingestion and indexing

3. Natural language search and retrieval

Model training and development

The process begins with building and training specialized AI models using both public and proprietary data sources to augment metadata. These systems can segment videos, identify and label specific actions (like hits or home runs in baseball, dunks or steals in basketball, or sentiment and type of scene in movies), and apply image recognition to isolate and identify actors (for movies) or players' faces and uniforms (for sports). The key is training and combining multiple machine learning models and data sources to achieve optimal results.

Content ingestion and indexing

Once models are trained, the system can be deployed to automatically process entire video libraries. For example, a baseball game can be automatically segmented into plate appearances and key plays, tagging players and actions, and tying those to the specific time or segment of video. This generates rich metadata that is stored in a database, creating the foundation for sophisticated queries and retrieval.

Natural language search and retrieval

The final component enables users to query archives using natural language through retrieval-augmented generation (RAG) techniques. For example, users can ask for "Aaron Judge's five longest home runs against the Red Sox in the last five years" and receive precise video clips that jump directly to the relevant moments.

The technical implementation of AI-powered media retrieval goes beyond algorithms and involves critical infrastructure considerations. The physical location of data and compute centers significantly affects the speed and reliability of content accessibility, particularly for AI inference where proximity is crucial. 

Modern distributed cloud platforms address this challenge by bringing compute and storage closer to users and content, thereby reducing latency.

To address these infrastructure and proximity challenges, FPT has partnered with Akamai, which is recognized as offering the world's most distributed network and a global cloud platform that extends to more than 4,400 edge points of presence (PoPs) in over 130 countries. 

This extensive infrastructure combines Akamai's established content delivery network (CDN) with core compute regions and edge computing capabilities, empowering businesses to deploy applications and data closer to end users. The result is better performance, scalability, and security.

There are three critical levers that optimize the performance of AI-powered media retrieval:

1. Proximity

2. Model efficiency

3. Hybrid infrastructure

Strategically locating infrastructure reduces network latency. Akamai's vast network of edge PoPs enables businesses to achieve this proximity, ensuring faster content delivery.

Using distilled models and embeddings allows CPUs to handle text embedding and re-ranking efficiently, while GPUs handle vision and high-fidelity video analysis. This optimization is crucial for maintaining performance without compromising on the quality of media retrieval.

Running GPU-intensive batch processes in core regions while distributing CPU-based inference closer to users at edge locations optimizes overall performance. Akamai's hybrid approach supports this by providing both core compute regions for intensive tasks and edge computing capabilities for real-time processing closer to the user.

AI-powered media retrieval unlocks a multitude of compelling business opportunities, including:

• Long-tail content monetization

• Streaming platform differentiation

• Broadcast enhancement

Archives transform from cost centers to value generators, so previously inaccessible footage becomes licensable assets for news organizations, content creators, and other media companies.

For streaming providers that are fighting subscriber churn, AI-powered search offers more engaging experiences. An action movie fan can request clips of the top 10 action scenes from the last five years, discovering content they might never have found through traditional browsing.

Networks and leagues can rapidly assemble clips to enhance live broadcasts, creating more dynamic and engaging viewing experiences without the need for manual editing overhead.

Business success is achieved through a significant reduction in manual tagging and indexing efforts. This reduction allows teams to focus on higher-value activities. Building on this efficiency, organizations can expect to see increased session time and user engagement as AI-powered search helps users discover more relevant content.

With increased engagement and session time comes increased loyalty. Streaming platforms and content providers can see decreased subscriber churn as their applications become stickier and more valuable to users. 

Perhaps most important, companies can unlock new revenue streams from previously inaccessible content, transforming archived material into monetizable assets.

Agentic AI represents the next stage in the evolution of intelligent media systems. While current platforms excel at search and retrieval, agentic AI can create specialized workflows that build upon these capabilities.

For example, an agentic system could compare a professional player's swing technique with an amateur's form under similar conditions, automatically fetching relevant clips from the database and creating personalized coaching content. This represents the convergence of archived professional content with user-generated material for entirely new applications.

The transformation of media archives from static storage repositories to dynamic, searchable assets represents a fundamental shift in how the industry approaches content value. Organizations that can quickly surface relevant content will create new revenue streams, enhance user engagement, and build competitive advantages in an increasingly crowded marketplace.

The technology exists today to make this transformation possible. The question isn't whether AI can revolutionize media retrieval — it's how quickly organizations will adopt these capabilities to unlock the hidden value in their archives.

Success will belong to those who can effectively combine advanced AI algorithms with distributed infrastructure and innovative business models, creating systems that not only store content but make every moment discoverable and monetizable.

By combining the scalability and performance of Akamai with the consultative approach and experience of FPT, organizations can significantly enhance their AI-powered media retrieval capabilities to ensure faster, more reliable, and scalable content delivery.

Are you interested in learning more about how AI is transforming media retrieval or how to modernize and monetize on Akamai Cloud with FPT? Watch this video. To speak with our media and entertainment experts, fill out our meeting form to start the conversation.