IBC 2025 proves it: AI works, but workflows hold it back

IBC 2025: AI isn’t the problem: workflows are

‍

IBC 2025 proved that AI has an integration problem, not an intelligence problem. Every booth had a demo; every panel had a take. But beneath the buzz was a sense of urgency, a recognition that the hype is giving way to a pressure to prove value. 

‍

The stumbling block isn’t the AI models themselves. It’s the difficulty of plugging them into complex, fragmented workflows that weren’t built for them. 

‍

The numbers back this up. Despite $30–40 billion in enterprise investment into GenAI, MIT research shows 95% of pilots deliver zero measurable return, while just 5% extract millions in value. 

‍

What separates the winners from the rest? 

It's how they architect their workflows. The successful 5% understand that AI integration is fundamentally a systems design challenge. Workflows, not models, determine whether a project becomes part of the 5% or plunges into the chasm of the 95%.

‍

‍

Nobody can do it all

‍

At IBC, the strongest stories weren’t about ripping and replacing infrastructure. They were about augmenting what already exists and creating a safer way to experiment with AI in production. 

‍

Without an orchestration tool, stacks quickly become unmanageable. Cloud services, AI point apps, and media-specific platforms pile up into a tangle of APIs, formats, and workflows that don’t fit together cleanly. That’s where pilots stall.

‍

We also saw proof that no single vendor can do it all, a shift from the old "platform wars" mentality to ecosystem thinking. Even the largest platforms only solve a piece of the puzzle. 

‍

With orchestration, the picture flips. Media companies get a framework to stitch tools together, run controlled experiments, and accelerate time-to-value—without betting everything on a single vendor.

‍

The show floor reflected this shift: AWS pairing cloud services with AI companies like Twelve Labs, Grass Valley opening up AMPP workflows, integrators leaning into low-code orchestration instead of heavy middleware. None of them claim to solve it all. They’re proving that ecosystems win and orchestration is what makes ecosystems work.

‍

Taken together, what we saw were signs of an ecosystem learning to scale together.

‍

The orchestration imperative

‍

There was also growing chatter around agent-to-agent (A2A) protocols, with Google taking the lead. But this emerging trend actually reinforces the orchestration thesis. Right now, orchestration still depends on APIs and that’s not changing. Even with agent-to-agent protocols like A2A or MCP, the underlying calls still happen through APIs. What’s shifting is the coordination layer: how those calls are managed, automated, and scaled. That’s why orchestration remains the foundation. When A2A matures, it won’t replace APIs, it will ride on top of them, reinforcing the need for strong orchestration frameworks. 

‍

The takeaway

‍

AI is ready. The models work. The difference between the 95% who stall and the 5% who scale comes down to one thing: turning a patchwork of tools into a coherent ecosystem. 

‍

That's why orchestration is becoming the core of modern media operations, the difference between AI pilots that fizzle and AI implementations that transform businesses.

‍

Get in touch

‍

Are you ready to see how orchestration can unlock AI’s value in your workflows? Let’s talk. We’ll walk you through how qibb helps media companies connect, integrate, and scale their ecosystems. 

‍

‍