--- title: "IBC 2026 and the speed limit of AI: why your data productisation stalls even with state-of-the-art media tech" description: "IBC takeaways: why productised data, human decision-making and simplicity decide whether agentic AI creates real value in media." date: "2026-10-27T09:00:00+00:00" category: "Strategy" excerpt: "IBC takeaways: why productised data, human decision-making and simplicity decide whether agentic AI creates real value in media." reading_time: "5 min 17 sec" image: "@/assets/uploads/1793182905528-mearra_ibc2026.jpg" image_alt: "IBC 2026 key visual with the slogan Create the Moment – Where media and entertainment come to life, and the IBC 2026 logo" path: "Insights > IBC 2026 and the speed limit of AI: why your data productisation stalls even with state-of-the-art media tech" --- A few weeks ago our Business Technology Strategist, **Maija Helminen**, participated in IBC (International Broadcasting Convention) in Amsterdam. These are her take-aways, greetings and epiphanies from that event. I spent three full days in Amsterdam, indulging myself in multiple panel discussions, exhibitions and dozens of meaningful discussions. There was a vast amount of smaller and bigger themes, but these are the three top ones in my opinion. - Experience over brand; the same piece of content is worth different to different people depending on the format and platform it reaches the correct audiences through. - Building stopped being the hard part. Now the real blocker is people and the decisions needed to scale agentic AI’s benefits and impacts beyond individuals or teams. - Agent speed is only worth something up to the pace a human can decide at, which makes simplicity the skill to learn. ## 1\. Experience over brand: The same piece of content is not worth the same to everyone The value of a piece of content shifts depending on who is experiencing it, where, and how. Live television barely registers with a lot of younger audiences - on a television that is. Put the same moment in a TikTok reel with a Minecraft animation split screen and suddenly it does. Same content, different value, because the experience changed. And the effects on individual attention span aside, that’s potentially bad news for anyone still only pricing by contact. Instead, the real gains can be made in reshaping what already exists to match demand as it shows up right now. One recording, understood properly the moment it is made, becomes a dozen products for a dozen audiences. This means that instead of the platforms, we need to focus on the data the content itself creates as it moves through people’s screens, speakers, and headphones. Data that needs to be defined as a product, with a name, an owner and a price. That is where data productisation comes in. Content, data or a capability becomes a product once it has a name, an owner, a clear definition, a price and a buyer. It sounds almost too simple, yet it fixes some stubborn problems. The recording nobody labelled well enough gets made again because it cannot be found. Two teams call the same thing by different names and end up with two versions of the truth. Nobody owns the asset, so nobody decides who it is for. And value that is never defined cannot be measured, which leaves decisions to gut feeling - or the loudest voice in the room. This matters even more once agents join the work. An agent can only use what it can understand, and it cannot walk over to a colleague to ask what a field really means. Data with a name, an owner and a definition is the entry ticket. Without it, the agent produces confident output on top of a mess, and a human is left to work out whether any of it can be trusted. ## 2\. Building stopped being the hard part A quick walk on the show floor is enough to make me believe that if you can think it, there is most likely a solution for it. Or there will be one soon enough. If you have the money, that is. But affording the latest innovation no longer guarantees a competitive edge. The real race is happening deep in the everyday of businesses, as organisations try to prepare, mould or even reinvent themselves in order to welcome in the AI era with all its glory and gore. To some, this looks like analysing current systems and operations, optimising processes or standardising data. Essentially, popping the hood of the current toolbox and seeing where this new Swiss-army knife could fit in. In reality, however, introducing agents has less to do with tech and more to do with people. AI is at the same time just a new tool businesses need to learn how to best utilise, and so much more than another solution. And with a tool that seemingly eliminates the waiting period, it becomes even more pressing to include the pause manually, by stopping to consider how and when we decide what should get built, how the outputs reach the right person, and how that person gets to say yes or no in time. This brings us back to the first point. Productised data gives that person something concrete to judge: an asset, an owner, a definition, an audience. Without it, a colleague is asked to approve something nobody can describe clearly, and the likely answers are a cautious no or a yes given on trust. Those colleagues are rarely technical. They are the editors, producers and sales leads who know what a piece of content is worth to whom. Their understanding makes a product definition right in the first place, and their yes or no decides whether an agent’s output ever becomes useful. In an industry full of non-technical experts, that is where agentic AI will be won or lost. ## 3\. Speed only counts up to the pace of the person deciding When generation seems instant, the first instinct is often to hand an agent every draft, every summary, every workflow, and call it efficiency because a person is not doing it anymore. It is the same reason you cannot make a factory more efficient by running the conveyor belt faster. When the speed surpasses the people at the other end, more automated output leads to more manual work, more mistakes, more chaos, and now the added tax of spent tokens. A system is only as fast as its slowest link, and right now that link is us. Even the flashiest agentic tool on the floor drew the line in the same place as everyone else: people. If we want the machine to analyse and execute while a human still decides, we have to look beyond the speed and quality of generation, and make that decision as informed, simple and scalable as possible. The real skill is finding the exact spots where an agent genuinely outpaces a person, and keeping the result simple enough that it still fits into the human decision that has to follow it. ## Putting it together Defining data as products gives people and agents the same picture of what exists, who owns it and who it is for. That shared picture lets a non-technical colleague judge an agent’s output quickly. And quick, confident decisions are what turn the speed of agents into something the business can actually use. Skip the first step and the other two get harder, which is why I would start there. Thinking what is the AI readiness of your organisation? Instead of guessing, [take the test](/ai-readiness).