Institutional change tends to come in reaction to a crisis. For example, the IRS didn't decide to answer the phones—the Senate forced it to in 1998 after hearings of taxpayer horror stories.
Institutions don't reform themselves, because reform requires someone to overcome thousands of hours of paperwork and a dozen layers of an org chart. No one person ever could manage it.
AI breaks the pattern. Consider what I work on: the Individual Master File, the database holding every American's tax records. It went live in 1962 and is plausibly the oldest software system in continuous use on Earth, according to Guinness World Records. It runs on millions of lines of IBM assembly code that a shrinking pool of programmers can read. Nobody kept it because they liked it; rebuilding simply exceeded any human team's capacity.
It doesn't anymore. One person and their agents can metabolize an entire bureaucracy—read every regulation, migrate every record, and clear every layer—and build something new. At the IRS, we're proving it: tiny in-house teams, no vendors, taxes moving to the cloud and onto your phone.
After automation, every institution can be remade in a new image: public transport, treasury departments, tax agencies, permitting, universities, everything. Not reformed at the margins—totally re-founded. And it won't take a crisis. Just one person who decides to put their mind to it.
