From Traditional Outlets to AI-Personalized Content

Media distribution has moved from broad editorial choices towards increasingly personalised feeds. As a data and AI entrepreneur, I am interested in what happens when the same system can both select content and generate it.

Social feeds first made the people we knew part of the entertainment. Influencers extended that relationship: a person could feel familiar to an audience far larger than any circle of friends. Algorithmic feeds then made selection less dependent on whom we followed.

Generative AI adds another possibility. The platform can produce material for a particular person, using what it has learned about their preferences. Selection and production start to become part of the same process.

For a founder, the distinction matters. A marketplace normally helps people find one another. If software can produce the thing being bought, the platform's relationship with its suppliers changes.

Consider two examples. A logo service might generate a design directly from a brief. A recruitment service still needs a person and an employer to reach an agreement. AI could help them understand their requirements and assess a match, while the people remain central to the transaction.

I would ask where a product's value comes from: producing an output, helping someone make a choice, or connecting people who could not easily find each other. The answer determines which data is useful and why a customer would return.

Would I value a piece of content in the same way if it were generated specifically for me? Relevance is useful, but part of the appeal of art and writing is encountering another person’s choices. I want to see how products account for that difference, rather than assuming that closer personalisation always means a better experience.

Cloud collaboration showed how a specific improvement to the way people work could give a new product a foothold. For AI products, I would look for a similarly concrete advantage: something a user can do more easily or a task they could not do before.

An incumbent may have customers, data and distribution. A new company has to offer a sufficient improvement to make switching worthwhile. Adding a model to an interface does not answer that question by itself.

These questions informed my investment in Tailor AI, founded by Vittorio Banfi, which creates personalised daily podcasts. It is a concrete way to explore how people use content assembled around their interests.