The Knowledge Left With the Analysts

When the model writes the analysis, the institutional knowledge lives in the model, not in a person who can explain it three different ways. That's the part of the London finance-analyst story nobody's asking about, and it's the part that actually matters.

The numbers are real and they're not a forecast. Search "finance analyst" in London and you'll find around 80 vacancies. Four years ago the same recruitment site listed more than 350. That's a 75%-plus collapse in one of the biggest banking hubs on earth, in exactly the roles defined by data processing and pattern recognition, which is where AI substitution is most direct and least reversible.

So the framing as a cautionary tale about workers caught in transition has the tense wrong. This isn't a horizon risk. It's a completed adjustment. The "ghosts of London's past" treatment reads as labor nostalgia, and nostalgia is the wrong response to a structural signal. The displacement is done. The consulting industry is, in many shops, still selling readiness for it.

Here's what got skipped. Those 270 missing analysts weren't just labor. They were where the reasoning lived. How a number gets sanity-checked, which exceptions matter, what a weird quarter actually means. When the headcount left, that judgment didn't evaporate. It migrated into the models and the prompts, and almost nobody in financial services governed that migration on the way out.

The displacement is done. The knowledge governance hasn't started.

I don't know how much of that institutional knowledge is recoverable if a model gets swapped, or whether anyone could reconstruct it. That gap is the live risk, and it's the one the vacancy chart doesn't show.

"Use whatever you want" turns into a mess fast enough with tooling. With the firm's actual analytical judgment, the mess is quieter and far more expensive.