I’ll be honest, after three whirlwind days meeting the people shaping AI in Silicon Valley and San Francisco as part of INMA Media Tech & AI Week, I came home with more questions than answers. 

I opened our two-day conference at KQED with those very questions. And over the course of the sessions, panels, and hallway debates, we explored them from every angle. Now that I’ve had time to digest it all, I want to share where I stand on each one — and what they might mean for the future of news and media.

Q: Will there be a human internet and an agentic Internet?

A: Yes. We’re already moving from search → answers → action. We’re in the middle phase now but quickly entering an era where agents act for us. The agentic Internet is coming, and we need to prepare. (I’m working on an Agentic AI master class now.)

Q: How do we start rebuilding media for bots?

A: By structuring and controlling our data. Stop thinking in “article pages.” Break content into small, structured nuggets. These will actually serve humans and agents as stories get “built” for personalised formats. It’s not glamorous work, but clean data is the fuel of AI and the foundation of survival in the agentic world.

Q: When and how do we atomise all our content?

A: We only publish a fraction of what we create. The agentic Web will surface the iceberg of content — not just what’s visible but what sits below. Think beyond what’s published: reporter notes, transcripts, interviews, internal discussions. Tools like Otter.ai are early examples of how latent content such as interview transcripts can become valuable data.

Q: How do we personalise content and format for humans while keeping shared experiences?

A: Some formatting, yes, but we don’t need hyper-personalisation for core journalism. Focus on the substance. The story and its truth stay the same, even if the presentation differs by audience and/or platform.

Q: Will distribution still be through Web sites and apps?

A: Maybe, but possibly not. Prepare for all scenarios. Scenario planning isn’t optional anymore; it’s basic business strategy.

Q: Will advertising revenue completely disappear?

A: Unlikely, but it will shift. Ad space in chat and answer engines is limited, yet if monetisation pressure rises, advertising may well pop up. Netflix is an example of this: once subscription-only, now ad-supported. Expect commerce to grow significantly as part of this mix.

Q: Should we block the bots or not?

A: Start by measuring and managing. Block bad actors. But remember: Not all bots are equal. This space will remain nuanced. Control matters more than a blanket “yes” or “no.”

Q: Will we need to master GEO for discoverability?

A: SEO may morph into GEO (Generative Engine Optimisation). Some are convinced. I’m not (yet). But we’ll need at least some free, structured content visible to answer engines for discovery.

Q: What do we optimise for?

A: Traffic is less relevant. Engagement helps but isn’t the full story. Answer engines optimise for utility. Our metrics must evolve toward the same north star: usefulness to the audience.

Q: Is there an opportunity for training data after all?

A: Yes but not through one-off deals. Models still need current, accurate information. That will come from collective licensing, not individual arrangements.

Q: Who will find a successful, scalable content-licensing solution?

A: Many are trying: Tollbit, ProRata, Microsoft, AWS. The most promising model right now is Really Simple Licensing (RSL), a standard that works across both the human and agentic Web. Business models are still forming, whether through revenue share or licensing and the foundation is being laid.

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .