Essay
The craft carries across
I started in the news business in 1989, when media meant ink on newsprint. There was no internet, no social media, no 24-hour news cycle. News was a physical product: newsprint on enormous spools, paste-up, pica rules, and a hard deadline set by when the press had to roll. You learned accuracy because a mistake was permanent. There was no editing a story after fifty thousand copies were on trucks.
Everything since has been a variation on the same work: figuring out what matters, getting it right, and putting it in front of the right audience.
Then I watched all of it change. Desktop publishing replaced paste-up. The web replaced the press run. Digital replaced film. I lived every one of those transitions as a working practitioner, not a spectator, and each time the same thing happened. A new tool arrived. The people who understood the craft adopted it and got faster. The people who only knew the old tooling got left behind.
AI is now doing to software what desktop publishing did to typesetting. It is collapsing the distance between the person with the idea and the finished product. I have seen this shape before. That is the perspective I bring to directing AI development, and it is why I am precise about what I do and do not do. I do not write code. I define what gets built and verify that it works.
That distinction matters. I am not a software engineer. I take a business problem, turn it into a specification an AI implementation partner can execute against, review what comes back, test it, and put it into production on systems real clients depend on. The specification says what the system must do, what it must never do, and how I will know it is correct, and it is written before anything is built. The work goes out in scoped increments rather than one large request. Nothing is accepted until it has been verified against a live system, with a dated backup and a documented rollback path in place first.
Why are my specifications any good? Not because I know the tooling. Because I know the business. Thirty-five years across newsrooms, publications, operations, and accounting is what tells me which requirements are right, which trade-offs are acceptable, and when something is subtly wrong even though it technically runs.
It is also why I self-host almost everything. When you have watched platforms come and go for more than thirty-five years, you get particular about owning your infrastructure rather than renting your business from a vendor who may pivot, raise prices, or disappear.
The craft carries across. The tooling is always temporary.
This essay is assembled from things Aaron has written elsewhere and is being edited into its final form. The feed will carry the revision.