Industry insights into Agentic AI

Words Verity Butler

I’m going to begin this by addressing the robot-shaped elephant in the room. I know people are really getting tired of talking and hearing about AI. In many respects, so am I.

But it’s important to remember that AI is not simply one single technical entity, and instead actually comes in a large (and growing) number of shapes and sizes. From generative to predictive, conversational to agentic, even Shrek would have to agree that this technical onion has many hidden layers.

Ogres aside, innovation in AI – and our understanding of how it can be used – is snowballing. In this instance, we are going to be looking at the agentic variant, which has already been making considerable waves across the broadcast and streaming tech sector.

One thing that quickly became clear to me when sitting down with industry experts to discuss this topic is that this sector in particular provides an excellent proving ground for what can be achieved with this widely misunderstood tool.

“But what actually is agentic AI?” I found myself asking each interviewee.

Maria Minaricova is the director of business development at Fetch.ai, and she distinguishes it clearly from the more commonly understood forms.

“Agency means autonomy,” she outlines. “The idea is that I am able to speak to my agent and give it commands, but then the agent goes out and tries to achieve that on my behalf. This is one of the biggest differences, as people often immediately think of GenAI when it comes to agentic. Generative AI generates things: text, images, etc. Agentic means to achieve goals on behalf of its owner.”

Greater autonomy is one of agentic’s many flashy advantages over its generative predecessors, which are often limited by the information contained in the datasets upon which models are trained. “This wave of agentic AI will become quite spectacular when it truly hits the major broadcasters,” emphasises Paras Chugh, who is an associate practice leader for HFS Research. “But they need to start moving on it now, as content is getting commoditised fast. Everyone can become a creator right now with the right open weight models and other tools, but the key differentiation is the one who holds the IP.”

Why now?

Industry buzzwords come and go so fast it’s difficult for even those of us constantly writing about them to keep up. Some might question whether agentic AI is just another breed of AI mania that will fade out into the background in a few years’ time – or whether it really is that next tech wonder to completely alter the world’s course.

Trevor Neal, CEO of RedSquid TV, highlights to us his own decades’ worth of experience in helping industries embrace transformative technology. “I’m probably best known for being one of the original team members behind the M Star Digital TV Group,” he begins. “You may never have heard of M Star, but it was launched during a time when the television industry was going through the change of analogue transmission to digital, and then from digital to smart TV.”

This period holds a mirror to what we face at present, with the market at the time moving from the old, big CRT televisions to flat panel. Like now, it also seemed that a lot of those companies were just adding bits and bobs to legacy TV set-ups without thinking long-term strategy.

“We came along,” Neal continues, “and started from a blank sheet of paper, designing a digital TV from scratch. As a result, we took over the world. Within five years, almost every television manufacturer you can think of was using M Star silicon in its products. We ended up with a very dominant worldwide market share simply because we started again. I think BBC iPlayer and YouTube were the first apps that we began to port.”

Neal believes we are now at the tip of a very similar-scale iceberg, citing another world-changing example that many of us can remember well. “There’s an analogy I would like to use – relating to the mobile phone industry back in 2006. The leaders at that stage were Nokia and Ericsson, and then two things happened. The first was that you were able to get high-bandwidth internet on the phone. The second was that the processor in the phone became very fast. The end result was the birth of smartphones.”

This allowed Apple and Google to suddenly usurp the number-one spot that had been defiantly held by the previous mobile behemoths for many years. And it was the data and processing power upgrades that were the culprits.

“Now if you fast-forward to the present, the same two things are happening again. Most people’s homes now have fibre or very high-bandwidth broadband, and the processors inside a television set have followed the same route that the smartphone went through.

“What that means is the television is going to run new forms of applications and entertainment that were not thought about five years ago. This is a time where we will see the smart TV era begin to run its course and what we call the ‘intelligent TV’ era will begin to rise.”

Paras Chugh recently co-wrote a report on the wider industry that took a similar view. It is called ‘Media leaders can explore the next growth wave of agentic AI to reclaim control’. The report claims that ‘broadcasters have become aggregators of leading IP and distribution platforms, but still struggle with profit margins’; however, ‘publicly reported data indicates that tech-native platforms such as Netflix have achieved operating margins in the 30% range and annual growth of over 20% in recent reporting periods’.

Chugh shares his thoughts on the origins of this issue: “In the 20th century, the media broadcasters were the people running the show. They had control over all three of the pillars – which I call the ‘media trifecta’. They had reach, engagement and monetisation. They had strong control over their distribution channels, they had all the ad spends attributed to them and they were not just running the monetisation but also the culture.”

The reference to the birth of smartphones is something Chugh also observes, but in a different light from Neal. “The first wave of disruption was the internet and smartphones,” he explains. “Both of these disrupted two pillars of that media trifecta. What the internet took was distribution, and as a result it took monetisation. The second pillar of the trifecta, which is engagement, collapsed with the advent of smartphones.”

Where does the value lie?

Both Chugh and Neal have made it clear: the industry is at a make-or-break point when it comes to harnessing agentic AI. Next, we need to understand the opportunities available and how to implement them strategically.

An innovative example arose in July 2026, when Fetch.ai and RedSquid TV announced the world’s first operator-grade agentic AI TV platform – combining Fetch’s AI technology with RedSquid’s end-to-end TV operating system. Founded in 2017, Fetch.ai is a decentralised, blockchain-based machine-learning platform that builds autonomous AI agents to automate everyday tasks and economic transactions.

“Some of our core products include ASI, which is our own, agentic LLM,” expands Minaricova. “It connects all the agents in the background and provides a user with their own personal AI. We also have agents that are registering in AgentVerse, which is a marketplace where anyone can build an agent, with its own tools and tech stacks. The agents can find each other, negotiate, transact and communicate.”

Like the approach back in the day with M Star, RedSquid is looking to rebuild television sets from the ground up, and partnerships with the likes of Fetch.ai demonstrate the company’s strides to reach that goal.

“There will be two types of applications that will emerge in the next few years,” Neal highlights. “There’ll be ones that make your life easier but that will cost the TV maker or the AI vendor. But then there’ll be applications which the user will like to use – but actually offer the AI vendor a chance to start monetising through the television.”

Unlike conventional smart TV platforms, RedSquid TV controls the complete software stack. This unique architecture with embedded edge-AI hooks enables Fetch to tailor its agentic platform specifically for RedSquid TV, allowing autonomous AI agents to become deeply integrated within the operating system itself. As edge AI capability advances, those agents will increasingly execute directly within the television, reducing latency, lowering cloud costs and enabling faster, more private and responsive experiences.

The result presents a fundamentally different kind of television. Rather than navigating apps and menus, viewers simply express an intent: “Plan a family movie night,” “book the holiday featured in this documentary,” “order everything I need to cook this recipe.”

Behind the scenes, intelligent AI agents discover content, compare services, coordinate smart home devices, manage subscriptions, complete purchases and orchestrate digital services all on behalf of the household – and that’s all through natural conversation with the television.

Minaricova also highlights the enormous opportunities this brings when it comes to personalisation. “In one household you could have several TVs, watched by young children, adults, seniors, etc. Everybody has slightly different preferences and is consuming content in different ways.”

Agentic AI handles this by unlocking a hyper-personalised viewing experience, which reacts and adapts autonomously to the varying preferences of different members of the household.

As well as personalisation, a core opportunity found in Chugh’s report is a sense of control being returned to broadcasters.

“Going forward,” he says, “IP-based content will be extremely valuable, as it provides soul to the content, which is currently missing in a lot of the AI slop we see across many distribution platforms at the moment. That would be the first important opportunity to grab as media broadcaster.

“I recently met the chief AI officer of Hasbro,” he adds. “And they had a really innovative use case, which they termed ‘behavioural licensing’.”

Hasbro owns a diverse portfolio of major IPs such as Transformers and Peppa Pig. It, like many others, has had to confront the challenge of people using AI to create fake content based off the IP, leaving it unable to monetise it.

“What they decided to do is complete the picture,” Chugh continues. “They recently partnered with ElevenLabs to create licensable content of the voices or even videos from its IPs, which means any brand can come along and license the use of them officially. That means the quality would not deteriorate, and the brand are therefore happy because they are then able to create IP-led content at scale.”

Another set of opportunities that emerged from Chugh’s research centred around content moderation and localisation.

“These were already important considerations long before agentic AI reached the market. All the major tech giants were already moderating social media content through analytics and tools already available. But, with agentic AI, this is only going to accelerate and start to become affordable for the smaller players in the market and local broadcasters too.”

He emphasises the possibilities this can present in terms of accessibility for viewers often sat at the bottom of a broadcaster’s priority list.

“In India, there are many AI models that are trying to become regional content players – because there’s no AI model that’s been trained across Marathi and Haryanvi, or some of these local languages in India – of which I believe there are over 30.”

Who’s really in control?

AI has been controversial from day dot. But the continued emergence of agentic opportunities brings with it chances of a fresh start when it comes to the pressure points around privacy, data and human involvement.

We’ve heard the term software as a service (SaaS) thrown around in the media-tech landscape for quite some time, but HFS recently coined a new spin on the old phrase. “We trademarked service as a software,” Chugh explains, “as opposed to the commonly used version. The idea is for human expertise to become the core of an organisation, while AI becomes the scaling agent or accelerator, which provides a value jump and can unlock the ‘DREAM’ framework’.”

Keeping humans in the loop is a common assertion when it comes to AI adoption. But Chugh offers an addendum to this take.

“I would say there needs to be a human at the end of the loop, where humans can govern these agents. They can close the loops which are started by the AI agents and also manage them throughout their lifecycle. What’s happening right now with pilots is that a lot of organisations are creating agents and running them throughout the enterprise, but nobody is monitoring them.”

Chugh breaks down the ‘DREAM framework’ in his report, where the ‘D’ is about defining your vision, while the ‘M’ is about measurable business outcomes. “This is so important,” he expands, “because a lot of CEOs think of AI as a magic wand that, once implemented, solves all of their workflow inefficiencies.”

Minaricova echoes Chugh’s opinion: “People should absolutely stay in the loop and make sure they don’t rely solely on AI. A way of looking at it is that humans have, over the years, become increasingly comfortable with autonomy. A good example is flight travel, which now largely relies on autopilot. Despite this, you always want to have that human there to make those critical decisions.”

Fears around privacy and personal data is one of AI’s most pervasive roadblocks. Neal emphasises why the partnership between RedSquid TV and Fetch.ai has come at such a key time, particularly due to these concerns.

“Telco operators like BT, Vodafone, Virgin Media – they all spent a fortune digging up roads to run fibre into our homes, and most people aren’t using all that capability. As new applications surface, that unused broadband pipe can start getting truly leveraged. That leads onto the idea of the whole chain beginning to come from broadband operators themselves, where they also supply the display; making sure that Wi-Fi and TV works perfectly with your router – and using AI in the background to help ensure that.”

If broadband providers took control of the TV platforms themselves, this could in theory allow more effective monetisation – which aligns neatly with the findings of Chugh’s HFS report.

“At the moment, smart TVs monetise largely through user data and home screen advertising,” Neal elaborates. “If the intelligent TVs were supplied by the broadband operators, they’re already monetising through your £20-a-month payment. That therefore means they are able to provide you a much better-quality service, while giving intelligent TVs the potential for the user to have a refreshingly clean viewing experience.”

Broadband providers around the world tend to be national companies. Neal believes that, if they sat at the start of the television-set chain, that would keep people’s data both in their television and the country they are watching it from.

“The privacy angle is really important because at the minute, every TV you buy on the high street is monetised by a foreign TV platform owner. You have no control over the data it collects about you, or any idea of where it goes.”

Is the industry ready?

Fears around human oversight and privacy aside, one of the biggest hurdles facing businesses when it comes to effective agentic AI rollout is actually the one already on the track.

“What we’re seeing across industries is that enterprises are getting stuck at the pilot stage,” says Chugh. “As a result, they’re unable to reach AI adoption at an enterprise-wide level, where the true value of AI gets unlocked.”

If there’s ever an industry that wrestles with legacy equipment and processes, it is broadcast.

“Only they themselves can solve the legacy infrastructure problem,” agrees Chugh. “They need to look at replacing that – but on top of this the outlook towards AI needs to be changed by the leadership. Right now, leaders think of it as a technology problem and hand it over to their CTOs. In reality, this needs to be shared across the organisation – and each department needs to prioritise it. AI isn’t just an overlay above your existing infrastructure, it’s a complete transformation to all the workflows and processes you have.”

Neal and his team’s work at RedSquid has only just begun, and as he did with M Star, he is starting to put a pen to that blank sheet of paper.

He says: “If you put the team that pioneered the smart TV movement with the team that’s beginning to make a real difference in agentic AI, what can you really create? Television goes in waves. We had analogue, we had digital, we had smart. Those eras come and go. The next era is going to be intelligent.”

There is certainly a sense that we are on the precipice of something great when it comes to agentic AI. Minaricova takes us back 20 years to the last time this happened, to demonstrate just how great this might be.

“Think back to the nineties and noughties when the internet came along and no one really understood what it could achieve – and how much it would totally transform our lives. Now, here we are 30 years later and the internet plays a central role in almost every area of our lives.

“But I like to think of the internet as two dimensional. It’s just information you search through. But agents offer that third dimension; where they go out, armed with information and goals from the owners, and then actively reach out to other agents to achieve those goals.”

Read the full report from HFS.

This article appeared in our IBC 2026 issue