Wanted: analyst programmer
The most valuable technology person in your business over the next five years will do one job that currently spans three or four job titles. They will work out what the business needs, decide how it should be built, and build it. There used to be a name for this person, and your business probably employed one.
The analyst programmer sat with the finance team in the morning, worked out what month-end actually did, and wrote the code that afternoon. They maintained the result for years, which kept them honest about the quality of their own decisions. The title was common in British businesses from the 1970s to the 1990s, and it existed for a plain commercial reason. Most companies could not justify paying two salaries for what one competent person could do.
The industry then spent thirty years dividing the job, and the division was right. Systems grew too large for one head. Specialists produce better analysis, better code and better testing than generalists, and a business analyst who challenges a developer's assumptions catches errors that one person marking their own homework does not.
What has changed is the tooling, and it reaches every role on that team. The research a product manager commissioned, the design work, the code itself: each has become dramatically cheaper to produce, because tools such as Anthropic's Claude Code, OpenAI's Codex and Microsoft's Copilot accept instructions written in something close to the language the problem is described in. The teams were built for a world where each of those activities needed a dedicated specialist, and that world is changing.
What remains scarce is knowing what to ask for: understanding the business need, knowing what a sound system looks like, and saying precisely what is required. That is one job, and it is the old one.
In my experience the person who can genuinely do this is rarer than the number who believe they can. Google's DORA research, drawing on nearly 5,000 technology professionals, found they amplify whatever an organisation already is, so a capable business gets faster while a struggling one speeds up its problems.
In your business the capability most likely already exists, held by two or three people. The developer who understands revenue, the analyst who reads the code, the architect who asks what the change does to productivity. No job title on your organisation chart describes what they can do, and you will discover who they were the week one of them resigns.
One understanding now costs five or six salaries. The most expensive tool licences run to perhaps £2,000 to £5,000 a person each year against an average UK developer salary of around £55,000, so the savings are material.
Wanted, once again: one person who understands the business and can build what it needs.