End Up Here · Watching and thinking
A response to Christopher Barnatt
Christopher Barnatt's new ExplainingComputers video is one of the clearest accounts I have yet seen of what an AI-shaped web might become.
The source
Barnatt's argument is not that the internet vanishes. It is that the familiar business and interface layers of the web may be bypassed as AI becomes the first point of contact.
In the video's structure, search and visiting are disrupted; agents become interfaces; synthetic and assisted content changes what is published; social media loses some of its centrality; text becomes less dominant; and the web moves into something beyond the Web 2.0 pattern of pages, feeds, platforms and advertising.
Watch “AI & the Future of the Web” on ExplainingComputers. This page is a response and interpretation, not a transcript.
What changes
If an agent can answer a question, compare sources, find a deal or complete a task, fewer people may open ten separate websites to do the work themselves. Websites still contain the underlying information, but the visible visit becomes less certain.
The browser becomes less important as an interface. A personal agent can call services, run software and carry a person's intention between systems. That could remove friction, but it makes agency, permission and accountability essential.
When machines can generate and summarise immense quantities of text, the web may become less text-centric and less dependent on social feeds. Human experience, video, sound, place and trusted relationships may matter more—not less.
Where our ideas meet
His technical diagnosis meets my social question. Barnatt describes the likely pressure on e-commerce, web-based SaaS, search and social media. End Up Here asks what those changes could release: more personal agency, more room for meaningful activity, and less life organised around selling attention.
The agent should be a servant, not a new landlord. If an agent mediates the web, it must remain answerable to the person whose life it helps. It should show sources, expose choices, preserve the ability to browse and refuse, and avoid quietly turning every preference into a platform's commercial asset.
Websites still have a future. A page can be a durable place, a voice, an archive, a body of evidence, a relationship and an invitation. The point is not to keep every old click alive. It is to make the human work worth finding, citing, visiting and returning to.
Beyond Web 2.0 need not mean beyond people. The hopeful version is a richer web in which agents handle tedious coordination while people retain curiosity, judgement, friendship, discovery and the pleasure of encountering something unexpected.
The open questions
Will agents widen access to the world's knowledge, or concentrate discovery inside a few proprietary assistants? If a creator's work is summarised without a visit, how are creators supported? If agents transact for us, who bears the cost when they misunderstand us?
These are not reasons to reject the future. They are design questions. The web's next layer should be open enough to inspect, plural enough to resist capture, and personal enough that a person can say: this is my agent, acting within my values, under my control.
Related End Up Here work
Making money with AI asks whether AI is the product or the leverage behind a useful ordinary proposition.
Governance asks how people might coordinate fairly when personal agents become powerful intermediaries.