AI agents: the next generation of engagement
Adam Lewis, CEO at AxiumAI, explores how operators are starting to look beyond traditional loyalty models and towards more dynamic ways of engaging players, supported by AI agents that can operate in the moment.
Sport is real-time – sportsbooks aren’t. That is where value is lost. AI agents close the gap. They combine live game context with real-time player behaviour to decide and act instantly, delivering the right prompt, narrative or opportunity at exactly the right moment.
Engagement becomes continuous. Context-aware. 1:1. This is not about better campaigns or richer rewards. It is about showing up at the moment of intent, keeping players excited, inspired and playing. Because in a real-time world, loyalty is earned in the moment.
Rather than rewarding behaviour after the event, engagement is starting to happen during it. Key moments in a match, a goal, a substitution, or a change in momentum, create natural opportunities to interact with players. Increasingly, operators are looking to align engagement with these moments, rather than relying only on more static, template-driven approaches.
AI agents have been built specifically for this environment. By combining live match understanding with player-level behavioural intelligence, these systems can deliver personalised engagement that reflects both the game and the individual. Content, insight and bet propositions are generated in real time, adapting continuously as the action unfolds.
This also changes how the sportsbook experience feels. Historically, it has been driven by menus and navigation, with players expected to find opportunities themselves. What is emerging instead is a more responsive experience, where insight and propositions are surfaced at the right moment, helping players understand what is happening and where to engage. Industry discussions increasingly point to this shift towards more interactive experiences that respond to player intent and live context.
That shift is being spearheaded by behavioural data. In the past, engagement was based more on broad segments and what players had done before, whereas now it can respond to what a player is doing at that very second, what they are looking at, how they interact, and how their preferences change during a session. This makes it possible to deliver more personalised, 1:1 engagement, where each interaction reflects the individual player rather than a wider group.
In practical terms, this might involve delivering a relevant proposition at a key in-play moment or providing insight that builds confidence in a selection. The focus is on making engagement feel connected to the individual and the game, rather than separate from it.
Commercial impact and continuous optimisation
Early deployments are already demonstrating clear commercial impact. Intelligence-led engagement, driven by real-time interpretation of live sport and player behaviour, is delivering consistent increases in both transaction volume and revenue. In several cases, operators are seeing double-digit uplift in activity alongside measurable gains in net gaming revenue per player.
Crucially, this shift is not just about doing more, but about doing it more efficiently. By replacing broad, incentive-heavy campaigns with precise, context-aware interactions, operators can reduce promotional spend while increasing topline revenue. The result is structurally improved contribution margins, particularly important in an environment of rising taxation and regulatory pressure.
Performance improvements are not isolated to single touchpoints. Operators are seeing more diverse bet participation, deeper in-play engagement, and stronger continuity across the full event lifecycle, from pre-match through to in-play and post-settlement. By connecting players with relevant content at the right moment, engagement becomes continuous rather than episodic, increasing session length, retention and overall player value.
At the core of this is continuous optimisation. These systems do not rely on static rules or predefined segments; they learn and adapt in real time. Every interaction feeds back into the model, refining future decisioning and compounding performance over time. This creates a dynamic system where engagement improves with scale and usage.
There are, however, important considerations as this capability evolves. Greater sophistication introduces the risk of complexity, particularly if experiences become less intuitive for players. Maintaining clarity and coherence in the user experience remains critical.
Equally, the application of behavioural intelligence must be handled carefully. The same data that enables highly personalised interaction also provides a deeper understanding of player behaviour, creating an opportunity, and responsibility, to support more informed, sustainable and responsible engagement strategies.
Operationally, this represents a fundamental shift. Engagement is moving away from scheduled campaigns and manual workflows towards continuous, real-time systems. AI agents interpret context, make decisions, and execute actions in parallel – coordinating across channels and moments rather than operating in isolation.
This marks a transition from fragmented, sequential processes to a unified control layer, where multiple agents operate simultaneously, share context, and act in real time. The outcome is not just more effective engagement, but a fundamentally different operating model – one that is faster, more adaptive, and structurally more aligned with how sport and player behaviour actually unfold.
AI-driven operating models
So, are loyalty programmes still fit for purpose? They continue to play a role, particularly in rewarding ongoing activity and supporting baseline retention. However, they are no longer the only way to drive engagement. Increasingly, they are being complemented by real-time, personalised engagement that operates across every touchpoint. Loyalty is no longer defined only by points or tiers. It is shaped by how relevant, timely and connected the experience feels throughout the player journey.
As AI agents become more embedded across sportsbook operations, they begin to form a more coordinated operating model, where risk, payments and customer experience are no longer siloed functions but part of a connected system. In this model, AI is no longer just informing decisions, but executing them, with agents capable of interpreting context, taking action, and collaborating with other systems without relying on manual orchestration.
So, with the industry continually evolving, the combination of structured loyalty and real-time, context-aware engagement is likely to play a major part in how operators interact with players and drive value moving forward.
Over time, this moves sportsbooks towards a more unified intelligence layer, where sustained interest and loyalty are not managed in isolation, but continuously optimised as part of a wider, connected ecosystem of AI agents.
