You have argued that understanding user needs is no longer the hardest part for media organizations. If the insight is already there, where do you see the real bottleneck today: in editorial culture, data architecture, product workflows or leadership?

Dmitry: All of those things that you mention, but the real bottleneck is not these teams individually – it’s the gap between them.

Most publishers I work with have more audience understanding than they know what to do with. They have research, analytics, user needs frameworks and product insights. What they don’t have is a shared operating model that connects those insights to actual commissioning decisions, format choices and distribution logic.

I’ve started describing this as our industry having solved understanding but not delivery. Editorial knows what audiences need. Product is building for a different brief. Data is measuring something neither of them defined. Leadership is holding three separate conversations and framing them as a strategy. That’s why, by the way, we have so many alignment meetings – because we are trying to smooth something over that was not designed to be properly connected.

The organizations making real progress have stopped treating editorial, product and data as separate functions that occasionally collaborate. They treat them as one audience system with shared language, shared taxonomy and shared objectives. That’s the real bottleneck.

The User Needs Model was originally designed to improve editorial decision-making. As AI increasingly influences commissioning, packaging and distribution, how can news organizations ensure those decisions remain driven by editorial intent rather than algorithmic optimization?

Editorial intent has to be explicit and well defined before AI can support it, as simple as that.

If your newsroom can’t describe why a story exists – what user need it serves, what format best delivers that need, for which audience at which moment – then AI has nothing meaningful to work with beyond engagement signals. And if you let it optimize for engagement signals alone, you’ll get more of what already performs. And more of the same won’t make you indispensable.

This is why I think the user needs framework became more valuable in an AI environment, not less. It gives us a human definition of value that can actually guide machine decisions. Instead of asking AI to maximize attention, you’re asking it to maximize your editorial mission.

User Needs shifted journalism from topic-led to intent-led publishing. AI will shift publishing from intent-led to system-led delivery of intent. The main challenge facing media is whether the intent being delivered is yours or the algorithm’s default.

In your view, what does an “AI-ready newsroom” actually look like beyond the technology itself? What needs to be in place – in terms of taxonomy, metadata, teams and governance – before AI can genuinely improve content strategy?

Much less exciting than people expect – and much more demanding.

Before any technology conversation, you need to know what you’re making and why. That means governed taxonomy connecting user needs, topics, formats, audiences and platforms. But taxonomy is the foundation, not the building.

An AI-ready newsroom has made a clear editorial decision about what only it can do – and what it shouldn’t bother doing. Commodity content gets automated. That frees journalists for original, differentiated work that no algorithm can replicate: the investigation, the narrative, the coverage that serves an audience need nobody else is meeting.

It also means moving from passive analytics to actionable editorial intelligence. Not dashboards that describe what happened – systems that suggest what to do next. Commissioning recommendations will be trained on your editorial values, arriving at the moment a journalist is deciding what to cover and how.

And every topic area should be delivering a digital service, not just content. AI makes that kind of service journalism scalable for the first time – but only if you’ve defined what service you’re actually providing – from politics and health to books and AI.

The AI-ready newsroom is not about the best models. It’s about being clear enough about what you are that AI can help you be better at it.

If there’s one idea you hope people take away from your session, what would it be?

Journalism needs to think of itself as an ecosystem. AI made content abundant. What becomes genuinely scarce is the ability to consistently deliver the right journalism, in the right format, to the right audience, at the right moment, everywhere they are. Individual great stories still matter. But they’re not enough on their own anymore. Your clarity on what you are and what you are not matters the most.

The question I’d leave everyone with is the one I ask every organization I work with: if you disappeared tomorrow, what would genuinely change in your audience’s lives? The answer to that question depends less and less on the quality of your journalism alone, and more and more on the infrastructure that delivers it. That’s what an indispensable newsroom is actually building.