The End of the Internet
The End of the Internet
Ben Thompson connects three shifts. Open publishing made content abundant, so he found a viable business serving a narrower paying audience. Public social feeds grew too crowded to feel like communities, while private group chats kept smaller groups useful. He argues that agent-made software will follow a similar path: rather than compete to distribute another public app, people can build for themselves and trusted colleagues.
He reports that Stratechery has made private tools for podcast production, task tracking and calendars, and that he has made tools for his home and sports viewing. He expects cheaper creation to uncover needs too small to justify traditional development, perhaps making many local markets rather than one giant software market. Security is a second reason he favors smaller audiences: his managed WordPress host patched a PHP flaw, and he expects AI-assisted attacks to make self-hosting in public harder.
The PHP maintainer confirms a security release, not the forecast that public hosting is finished or private apps are safe. Thompson provides no code, repeat-use, permissions, maintenance or total-cost record for these tools. A separate internal-tool operator account in the linked notes describes a much longer parallel run before a private agent-built tool replaced spreadsheets. Smallness may change distribution and exposure; it does not by itself establish trustworthy shared software.