In our recent analysis ahead of CES 2026, we assessed how generative video and synthetic talent are transitioning from experimental applications to production infrastructure, reshaping the way media companies create content.
This second piece provides further insights into the trends emerging on how audiences discover and experience that content as we move into 2026.
AI-driven personalisation in media has been maturing rapidly, with recommendation engines now determining how the majority of content reaches viewers. However, this coming year will see this evolve further: from passive recommendation systems to active agentic curation. Below, we delve into this in more detail and explore the architectural demands this will put on companies looking to deploy these capabilities at scale.
Personalisation: the evolution to date
Personalised audience experiences have become foundational to streaming platforms. And the evolution continues. Spotify’s introduction of “Prompted Playlists” last week exemplifies this. The new feature enables users to generate highly personalised playlists based on the full arc of their listening history through natural language prompts.
Unlike the streaming platform’s earlier recommendation features, these playlists take into account years of listening data, global knowledge, and scheduling preferences, allowing users to request specific content and receive tailored results that refresh on their chosen schedule.
The strategic impact of sophisticated personalisation engines is notably significant. Netflix attributed 75% of content watched on its platform to AI-driven recommendations, a system the company reports saves approximately $1 billion annually by reducing subscriber churn and improving content utilisation. These features, which were once a bonus, have become the core infrastructure determining which content reaches audiences and how media libraries generate value.
The move to conversational recommendation systems
However, while these systems are now well established, 2026 signals a step beyond traditional personalisation engines. Media discovery is beginning to shift toward more interactive discovery models. And a number of major media companies are coming to us to talk about how this could be productionised. Netflix’s research into conversational recommendation systems (CRS) reinforces this direction. Their academic work demonstrates efforts to integrate collaborative filtering with natural language understanding, enabling users to describe preferences conversationally rather than navigating visual interfaces. It’s still relatively early stages but we expect much bigger discussions on this at CES.
Agentic AI as an enabling layer
As these conversational systems mature, agentic AI is becoming an enabling layer for next-generation content discovery. Agentic recommendation systems represent a next step in how personalisation can function, replacing reactive pattern-matching with models that anticipate needs and drive proactive curation.
While e-commerce has led adoption of these approaches, demonstrating the ability to proactively surface content and anticipate user needs across channels through AI agents, we expect to see media and entertainment increasingly explore similar principles as we move into the new year and we look forward to progressing these activities.
In the context of streaming and on-demand media, agentic approaches will enable systems to independently orchestrate the viewing experience – interpreting viewing patterns, contextual signals, and stated preferences to surface relevant content without requiring users to initiate searches or browse categories.
The new technical demands for media and entertainment
Going forward, conversational and agentic discovery will require systems able to operate autonomously in real time with minimal latency. This demands infrastructure capable of delivering personalised curation at scale across millions of concurrent users.
These developments require media organisations to recalibrate technical capabilities and re-think legacy systems completely, ensuring they are instead architecting for rapid adaptation, flexibility and embedded governance. Retrofitting capabilities as market conditions evolve is no longer an option.
Beyond technical implementation, questions of transparency, control, and accountability will become increasingly central. And in our next article within this series, we will explore these architectural considerations in further depth.
The media organisations succeeding in 2026 and beyond will be those that understand that AI development is fundamentally an architectural task. As CES brings these capabilities into sharper focus, building the resilient foundations that make these capabilities governable, secure, and adaptable over time must emerge as the strategic priority.
