2026 Accelerator Project

Project Objectives

Champions: ITV, AWS, TCS, Astro Malaysia 

Participants: Camb.ai, Admongrel, Shure, Dalet 

This project aimed to explore how agentic AI could be applied across the full live media value chain, from production through to distribution and consumption. 

It focused on orchestrating multiple AI-driven capabilities, including real-time content understanding, automated production tasks, highlight generation, localisation, personalisation, and monetisation. 

The objective was to move beyond isolated AI use cases and demonstrate how coordinated, end-to-end workflows could transform live content into dynamic, multi-platform, and personalised experiences. 

This included enabling “Me-Cast” scenarios, where content was continuously adapted to individual viewers, while remaining grounded in practical production and delivery environments. 

Innovation and Collaboration 

Across the industry, AI had largely been applied in isolated use cases, often as standalone tools within either production or distribution workflows. Meanwhile, live media operations remained complex, manual, and siloed, limiting the ability to fully leverage real-time content. There was a need to move towards integrated, agentic approaches that operated across both production and distribution, enabling content to be created, adapted, and delivered as part of a continuous, intelligent workflow. 

This included applying AI within production to automate and augment tasks such as logging, segmentation, highlight creation, and editorial assistance, while simultaneously enabling downstream capabilities such as localisation, personalisation, audience engagement, and targeted monetisation. These capabilities needed to be orchestrated together, allowing content to flow seamlessly from live production into multiple tailored outputs in real time. These approaches were then tested within production and distribution environments to understand how they could be applied in practice at scale. 

Broadcasters and content owners defined production workflows, editorial requirements, and rights constraints; technology partners contributed agentic systems and AI capabilities across the chain; and platform partners supported distribution, engagement, and monetisation models. Working together enabled the development and validation of end-to-end, multi-agent workflows that were both operationally viable and scalable across the industry.

Proof of Concept Demonstrated at IBC: 

At IBC2026, the project demonstrated a working agentic AI environment for live sports, bringing together multiple agentic capabilities within a shared live-media workflow. 

The proof of concept showed how specialised agents could operate across different stages of the live content chain, including semantic understanding and automated highlight creation, multilingual localisation and dubbing, live content moderation, personalised advertising and monetisation, and customisable audio experiences.

These capabilities were connected through an orchestration layer designed to coordinate agents, share context and enable different outputs from the same live content.   

The demonstrations included live content being analysed and enriched to create context-aware highlights; commentary being translated and dubbed into multiple languages; content being identified and remediated for compliance while still live; advertisements being selected for individual viewers; and personalised audio experiences built from isolated and enhanced live sound.

Human oversight and responsible AI guardrails were incorporated into the workflows, allowing editorial teams to review, approve or block AI-generated outputs.   

Live at the IBC pod, Shure and Camb.ai also demonstrated a live workflow of audio capture through the localisation agent.

Project PoCs:

View the 2027 Proof of Concept assets the team of AI for Live Sports and Beyond: Rewiring live media via Agentic AI Orchestration and “meCast” presented at IBC2026.

View the Project's PoC Presentation 

Key Learnings: 

The proof of concept showed that agentic AI can connect multiple stages of the live media workflow, but successful orchestration depends on more than simply connecting individual AI capabilities. Agents need access to shared content and context, clear roles within the workflow, and an orchestration layer capable of coordinating when and how each capability is used.  

Working in a live environment also highlighted the importance of latency, performance and consistency. AI capabilities need to operate within the time constraints of live production while maintaining sufficient quality and accuracy under production load. Cost and ROI also become increasingly important as these workflows scale.   

The project reinforced that human oversight remains essential. Rather than removing editorial judgement, agentic workflows can automate repetitive processes and surface decisions or content for review, while people retain control over higher-risk or uncertain outcomes. Responsible AI therefore needs to be designed into the orchestration itself, with appropriate controls around authenticity, safety, rights, consent, transparency and auditability.   

Finally, the project demonstrated the value of a modular, platform-agnostic approach. Different models, agents and technology partners can be selected for different tasks, allowing workflows to evolve as AI capabilities, production requirements and commercial models change. 

Industry Impact: 

The project provided a practical demonstration of how agentic AI could change the role of AI within live media: moving from isolated tools and individual automation tasks towards coordinated systems that operate across the content lifecycle. 

For broadcasters and content owners, this creates the potential to reduce manual and fragmented workflows while enabling content to be understood, adapted and distributed at live speed.  

For audiences, the same underlying live event can increasingly support experiences tailored by language, interests, audio preferences, and context. For commercial teams, richer understanding of both the content and viewer creates opportunities for more contextual and personalised forms of advertising and sponsorship.  

The wider significance is that the approach is not limited to sport. The architecture developed through the project provides a foundation that could be extended across news, entertainment and other forms of live media, with modular agents added or replaced as technologies and business requirements evolve.