Turning real-world footage into gameplay: inside the CCTV Game concept
A chance idea, inspired by CCTV footage and movement-tracking technology, has evolved into a new category of live content. Sam Jones, CEO of 155.io, explains how the studio is combining real-world video feeds with AI-driven analysis to turn everyday activity into structured betting markets.
Sam, where did the original idea for CCTV Game come from, and when was the eureka moment you believed the concept was practically/technically achievable?
Well, for the last two years we’ve been building alternative live content for players – including marble racing, plastic duck racing, Plinko games, coin flip games, and things like that. We spent a long time developing this in our own studio, and it’s obviously very operationally heavy running a 24/7 setup, as well as designing and manufacturing games with camera sequences and all that comes with it.
So part of me started thinking – is it possible to outsource some of this? We’d looked at bringing in third-party content before, but were underwhelmed by how it looked.
Then I got a call from a client asking if we could launch a game based on live CCTV footage of monkeys, with betting on what they do at different points in the day. While I thought that would be wrong from an animal welfare perspective, it did get me thinking about CCTV footage more broadly. That was really the spark about six months ago – could we create games based on real-life events, just not involving animals in cages?
The second piece came when I was at the G2E conference in Vegas last year. I came across a company using technology to track human movement around casinos – analysing pathways and behaviour so venues can optimise layouts and merchandising. It was quite similar to the kind of tracking technology used by Tesla to monitor cars. That led me to the idea that we could combine CCTV footage with this type of software to track, say, cars, and turn that into a betting market.
On the surface CCTV Game looks simple, but there’s clearly a lot going on underneath. How does the game function from a technical point of view?
In terms of how CCTV functions, what we do is take live CCTV footage either from partners who have licensed public feeds, or from our own cameras in different locations. So the source of the content is always real-world – none of it is AI-generated, it’s all genuine footage.
We then use artificial intelligence to take segments of that footage. So it’s not truly live in the moment, but it is taken from live feeds. We go back a few minutes and capture a 30-second snapshot of traffic, and the AI calculates how many cars pass through a defined tracking zone.
From there, we set the odds – typically an over/under range as well as an exact number – based on where the true count sits. A new game starts every 30 seconds, and we continuously rotate locations, so players never know what the next setting will be. They also don’t know which 30-second window is being used, and we’re constantly adding new locations into the mix.
If, for any reason, traffic stops – for example, if someone tried to interfere to influence the count – the AI simply skips that location. If there’s no movement, it’s removed from play.
For that reason, we believe it’s a very difficult game to manipulate.
CCTV Game blends AI, camera feeds and gameplay. How much of that balance was planned from the start versus figured out along the way?
In terms of how much of the balance was planned versus figured out along the way, I think the core concept was always there, but how it came together evolved quite naturally. From the outset, one of the key advantages we saw was the ability to offer highly varied, global content without needing our own teams and cameras in every location. From a player perspective, that creates something that feels very different – almost edgy, slightly dystopian, and very authentic. It stands out in a casino lobby because it doesn’t look or behave like traditional content.
From our side, it’s also far less operationally intensive compared to some of the live content we’ve built previously, which is a big shift. When it comes to AI, that piece became clearer as we developed the concept. The AI essentially acts as the referee – it’s counting the cars, monitoring the footage, and helping to generate the odds. So it’s not just a bolt-on; it’s central to ensuring fairness and consistency, while still keeping the mechanics relatively lightweight from an operational standpoint. On the gameplay side, we made a conscious decision to keep things very simple. The interface is deliberately stripped back, with just a small number of options for players to choose from. That said, we’re continuing to evolve this, and you’ll see additional features coming in over time.
The social layer is also very important. Players can see what others are predicting, who’s winning, and how much is being staked on each scenario, which adds another level of engagement. So really, it’s the combination of globally sourced real-world footage, AI acting as a referee, and a simple, accessible user experience that makes the product work
Streamers have picked the game up quickly – in fact it came up on my personal Instagram feed a few weeks ago. Was that social capture part of the plan, or did it happen naturally once it went live?
We’ve got some great partners across the business, including Roobet, Stake, Gamdom and Shuffle, and I think that’s been a big part of why the game has caught the attention of streamers. But it’s also down to the content itself – it’s just so different. No one has really created games based on CCTV footage before. From a streaming perspective, it can be quite boring showing slot or crash games; they’re not always that engaging to watch. CCTV, on the other hand, is genuinely unique – slightly crazy, even – and it creates something far more interesting and unpredictable for viewers. I think that’s a big part of why it’s taken off on social media.
You’ve just launched Snow Run, moving the same core mechanic into a completely different environment. What made skiing the right next step after Rush Hour?
We were a few weeks into the CCTV game gaining real traction in the market, and I was actually in Switzerland for some meetings. Being based in Dubai, I ended up getting stuck there much longer than planned – I spent about nine days there due to the war and the challenges flying back into Dubai with airport closures. Making the most of my stop, I spent some time up in the mountains and started looking at the kind of content available from ski lifts, slope-side cameras, drones, and things like that. That’s really where the idea for the second version of the CCTV game came from, which was Snow Run. So this is more of a hybrid approach. In Snow Run, we’re using publicly available camera feeds, some drone footage, and also POV (point-of-view) footage from people skiing and snowboarding down the mountain. Some of this isn’t live, as it’s obviously much harder to capture that kind of content in real time, but we felt it added an interesting dimension to a snow-themed game.
What’s different with Snow Run compared to the CCTV game is that it focuses on varying angles of the mountain, with slightly different gameplay mechanics. We felt it was a really strong addition to the overall product family. We also have the Duck River game, which uses multiple CCTV cameras on a plastic duck river, and we’ve got a few more concepts in development under the same brand.
Snow Run leans heavily on real-world footage – GoPros, drones, CCTV, creator content. How do you decide what environments actually translate into a game, rather than just interesting footage?
For Snow Run we took a few days to explore how we could capture different angles – whether that’s drones, POV cameras, ski lifts, or even people queuing to get onto the mountain. We tested a wide range of formats.
What makes the game interesting is that, within a 30-second window, there’s a lot of chaos and activity. Because of that, there were a few things we decided not to go with. For example, ski lifts are too automated and slow, and when people are standing in queues, there just isn’t enough variation. Some of the ski lift footage also didn’t offer enough fluctuation in numbers to make it engaging.
So we moved away from those initial ideas and focused more on drone footage, slope-side cameras, and POV content, which is much faster-moving and creates a far more engaging experience for players – especially when they’re watching people coming down the mountain and counting how many pass through. We tested a lot of formats and ultimately selected the ones that worked best for the game.
How do you ensure that remains consistent and trusted as you scale across different locations and footage types?
Ultimately, the plan is to have these games certified – both in terms of how we source the content and how the AI evaluates it. More broadly, the goal is for the CCTV brand itself to be certified, so players can fully trust what they’re engaging with. In the long run, that’s really the only way to build total trust in the market.
Snow Run is launching exclusively with Roobet before a wider rollout. What does that launch strategy allow with regard to testing and mean in terms of momentum?
As a result of Roobet’s belief in the product and promoting the first CCTV game, we agreed to give them an exclusive period for Snow Run. This allowed them to invest in streamers and promote the game across their network. The game is now available for general release and accessible to all other operators.
