AWS re:Invent 2025 - How Cires21 revolutionized advanced video workflows using AI on AWS (SMB203)

Summary of AWS re:Invent 2025 Presentation: "How Cires21 revolutionized advanced video workflows using AI on AWS"

Overview

• Cires21, a Spanish live streaming service provider, has developed a unified platform called "Media Copilot" to streamline and optimize video workflows for their media and entertainment customers.
• The platform leverages various AWS services and AI capabilities to automate and enhance content production, delivery, and monetization.
• Key focus areas include content personalization, semantic understanding, and real-time processing for live events.

Challenges Faced by Cires21's Customers

• Fragmented ecosystem of applications for media operations, leading to low content delivery, high costs, and duplicated work.
• Need for a more integrated, efficient, and scalable solution to manage the complexity of modern video workflows.

The Media Copilot Platform

Architecture

• Serverless architecture using AWS Lambda, API Gateway, Step Functions, and SageMaker.
• Leverages AWS media services like MediaConvert, MediaLive, and MediaPackage for transcoding, live streaming, and content delivery.
• Integrates custom AI models for tasks like automatic speech recognition, voice activity detection, and diarization.
• Uses Amazon Bedrock to generate advanced metadata like subtitles, highlights, and summaries.

Key Features

• Unified platform that integrates various AI-powered capabilities for media processing and workflow automation.
• Faster content delivery and lower costs compared to fragmented solutions.
• API-first approach allows seamless integration with clients' existing systems and workflows.

Agent-based Approach

• Developed custom agents using AWS Agent Core, a serverless runtime for building intelligent agents.
• Agents can perform complex tasks like finding viral moments, creating social media clips, and adding captions automatically.
• Leverages Agent Core services like runtime, gateway, identity, observability, and memory to ensure scalability, security, and context-awareness.

Lessons Learned and Future Developments

• Importance of selecting the right SageMaker deployment model (synchronous vs. real-time) based on the use case.
• Benefits of segmented video processing to enable parallel processing and reduce inference times.
• Plan to integrate more visual AI models to add more context to the agent-based workflows.
• Focus on developing specialized AI agents with narrower, more efficient capabilities.
• Prioritizing real-time processing for live events to enable immediate decision-making and content optimization.

Business Impact and Use Cases

• Enables media and entertainment customers to streamline their video workflows, reduce costs, and improve content delivery.
• Facilitates content personalization and targeted monetization through data-driven insights and AI-powered automation.
• Enhances the viewer experience by generating personalized content, multilingual support, and real-time highlights and summaries.
• Improves operational efficiency and agility for live event coverage and social media content creation.

Conclusion

Cires21's Media Copilot platform demonstrates how media and entertainment companies can leverage AWS services and AI capabilities to revolutionize their video workflows. By addressing key challenges around fragmentation, efficiency, and personalization, the platform enables customers to deliver better content experiences, optimize operations, and unlock new monetization opportunities.