Pour yourself a stiff drink because media production is about to get wilder than Quentin Tarantino's foot fetish.

Multi-agent workflows (or "agentic workflows") is a topic on the tongues of both Silicon Valley developers and cinema screen directors. This approach to overcoming the limitations of current large language models marks an advancement in the technology and poses a potential bridge to artificial general intelligence.

What are multi-agent workflows?

Multi-agent AI workflows involve several AI systems (agents) working together to accomplish tasks that are complex or too large for a single AI to handle. Think of it like a team where each member has a specific skill set: one might be great at planning, another at design, and another at organising information. By working together, they can achieve goals more efficiently than if they were working alone. Prominent AI figure Andrew Ng has noted that such workflows don’t just create efficiencies, but rather fundamentally alter the interplay between technology and human creativity.

In the context of media production, this can be incredibly powerful. For example, in creating a movie, one AI agent might analyse scripts to suggest plot improvements, another could generate realistic visual effects, while a third manages the project timeline and resources. Such a workflow is an iterative process in which the agents refine their outputs through collaboration and self-improvement. That is, the media output continues to be optimised as more information is received.

Why is this significant?

If the magnitude of change this brings to production methods isn’t immediately apparent, imagine a future where a lone director, armed with an arsenal of AI agents (let’s call them “SpielBots”), can oversee an entire Hollywood production process.
As these AI systems take on roles ranging from scriptwriting and editing to visual effects and project management, the traditional roles of human crew members are fundamentally questioned. This transformation not only challenges the existing dynamics of media production teams but also raises critical concerns about the future of employment, skill development, and the human element in creative industries.

Practical applications in media production

The value of agentic AI workflows, especially in fields like media production, lies in their potential to revolutionise how content is created, distributed, and consumed. Here are a few key reasons why this is important:

• Efficiency and Speed: By dividing tasks among specialised AI agents, projects can be completed faster and more efficiently. This means content can be produced and updated at a pace that matches the rapid consumption habits of modern audiences.
  
  
• Creativity and Innovation: Multi-agent AI can push the boundaries of creativity, offering new perspectives and ideas that might not be immediately obvious to human creators. This could lead to more innovative and engaging content that captures audiences in new ways.
  
  
• Personalisation: AI systems can analyse vast amounts of data to understand audience preferences and trends, allowing for highly personalised content. This not only enhances viewer engagement but can also help content creators target their audiences more effectively.
  
  
• Cost Reduction: Automating parts of the production process can significantly reduce costs associated with manual labour, rework, and time. This can make media production more accessible to smaller creators and studios, democratising the field.

Case Study: Inworld

In a recent demonstration by Inworld, cofounder Kylan Gibbs demonstrated the potential of agentic AI workflows to transform game development and player experience. In a collaboration with industry leaders, Inworld introduced an AI-powered engine that seamlessly integrates with development platforms such as Unreal and Unity. This engine revolutionises real-time in-game experiences by dynamically responding to game states, thereby enhancing narrative depth and player engagement. Through a demo, Gibbs illustrated how AI-driven conversations and scenarios could surpass traditional gameplay mechanics, offering a more immersive and personalised gaming experience.

This innovative approach presents the engine's ability to generate dynamic game states and quests based on player interactions. Using a network of AI models, Inworld AI’s platform allows for real-time adaptations in gameplay, creating a rich, responsive environment where player decisions dictate outcomes. This not only showcases the engine's versatility in crafting engaging narratives but also highlights the broader relevance of multi-agent AI in media production. By enabling more interactive and immersive storytelling, this technology paves the way for future advancements in entertainment, potentially extending its benefits to the production of films and other media.

How Can This Be used in filmmaking?

Just as Inworld AI's engine dynamically alters game states and narratives in response to player choices, a similar AI framework could be employed to enhance film production processes, character development, and narrative complexities. In such an application, AI agents could collaborate to analyse scripts in real-time, suggesting alterations that heighten emotional impact, deepen character arcs, or better align with audience expectations based on vast datasets of viewer preferences and behaviours.

This technology could also extend its influence beyond the script, assisting in directing, where AI could suggest camera angles, lighting setups, and even provide real-time feedback during shoots based on the envisioned narrative outcome. For post-production, AI agents specialised in editing could offer recommendations for cuts or transitions that enhance the storytelling or viewer engagement, leveraging real-time audience feedback from early screenings.

The Final Cut

Multi-agent workflows promise a new horizon where technology and human creativity fuse to produce more captivating content. In many respects, such workflows echo the collaborative and iterative nature of filmmaking, albeit lightyears faster than even Netflix can produce a season of Too Hot To Handle. Regardless, it's the dawn of a new era in narrative storytelling, one in which the line between creator and creation blurs.