Death Throes, Take Two

Big organizations have a posture they take when a wave shows up that will rearrange their industry. I saw it once from the inside, during the last wave, and I think I'm seeing it again right now with AI.

Let's look at how it shows up today, in one person's work week, and then at what else is doing it.

Death throes

"Death throes - the violent fits of a beast that doesn't know it's dead yet."

When a real wave comes through, the biggest orgs don't adapt and they don't die. They flail. They respond hard - committees, new org charts, vendors banding together - and from the inside all of it feels like leadership. From the outside it looks like a lot of meetings and a lot of vendors, without much change in what people actually do.

I have a specimen this time. I've seen this same gap from the inside.

A support team, right now

Say Maya runs a customer support team at a big company. How much AI does her company use? Depends who's counting. Let's walk her week, step by step.

Read (b) next to (c). I think the specimen lives in that gap. The measurement seems to say almost nobody uses the stuff. The work looks like the team runs on it. Maya isn't an early adopter and she isn't resisting. She's just doing her job, and the job changed underneath the reporting. The dashboard and the work are measuring two different jobs.

Bentonville, 2004

I joined Walmart in 2004 as a programmer analyst, three years after the dot-com crash. The crash had made the skeptics look right, and meanwhile the actual wave kept building underneath, because e-commerce was probably coming no matter what anyone's feelings about the stock market were doing.

By the end of my run there I was Head of E-Commerce Marketing. The channel I inherited was a fraction of total revenue, and it was already influencing the majority of buying decisions. People mostly did the deciding online and then the buying in a store. That's probably why Maya looks familiar to me, I think. The fraction and the majority seem to be the same two facts. The books measured the fraction, and the decisions stayed mostly online.

Where the throes show up

Alright. Let's get above it. Two things in the news and in conversation make my gut say death throes.

First, the cartels. Big companies seem to be banding together into groups and panels to set policy and standards, while hardly anyone inside those companies uses the tools day to day. It's WAY early. Deciding anything durable right now probably takes years of ordinary use that nobody has logged yet. A panel can't wait that long, so it produces frameworks and statements instead. The real answers - the ones that only come from unglamorous, everyday use - will show up later, on their own schedule.

Second, the split inside the big shops. Some people say rules first, adoption second. Others say yolo, we're going for adoption. Either way, the maneuvering has started. The ones shouting loudest about readiness frameworks read to me like an org chart bracing for a wave it claims it already caught.

Not every one of those groups is theater. Some of them do real work. But the two news items seem to share one shape, I think. Both of them have a lot of structure getting built at the top, while the everyday use underneath stays thin.

The pattern, off the specimen

Let's isolate the death throes so we can use it elsewhere.

Take lending. Rates and policy do what they do, and certain well-known lenders answer with structural gymnastics. Who's solvent and who isn't, I have no idea. That isn't advice. We're hunting meta. The exercise that seems useful is watching which lenders respond to rate pressure by reorganizing the committee, and which ones quietly rebuild the plumbing. The committee version is motion that resists the change. It rearranges titles and reporting lines while the actual work of moving money stays about as hard as it was. The plumbing version changes how a loan gets made at all.

Accounting wears the same posture with different furniture. Software eats the bottom of the work first: the data entry, the reconciliation, the first draft of the return. The big end of town answers with standards panels and credentialing moats, while the small firms just use the software. That defensive architecture says the work can only come in through licensed hands, at the approved pace, under the approved letterhead. The moat guards the entrance. But the tools came in through the staff's laptops, not the entrance. The bottom of the work is already being done while the panel is still wording its statement. That's my read, anyway.

There's also a principle in here I didn't expect. Full adoption was maybe never the point. In my retail years, e-commerce stayed a fraction of revenue for a long time while it took over the decisions. Winning the decisive share of attention was enough. Being an important enabler - and never the majority line item - carried the whole thing anyway. If that holds for AI, then every org measuring adoption against 100% is probably grading the wrong exam.

One note against myself, honestly. There's a real chance that seeing throes everywhere is just me matching new shapes to an old biography. I lived through one wave, so now every flail looks familiar, and the Bentonville years are the reference I reach for whether the fit is real or not. One wave is a small sample. (I know.)

What I think I know from that stretch: the loudest organizational motion shows up right at the inflection, and it seems to point against the current. I think the committees and cartels are mostly resisting, tbh. The wave keeps moving anyway, as far as I can tell.

The open question is what happens when the convulsions get big enough to move something real. I don't mean meetings. I mean the law, the compliance burden, the list of vendors a hospital system is even allowed to buy from. If one of those panels writes a rule next year that decides what Maya's company may deploy to her team, does her company's flailing stop being a symptom and start being the weather she has to work in? I don't know. I keep circling that one.