The most consequential shift in software right now is not a new framework. It is that producing a plausible first draft of a component has become close to free. That single economic change moves the bottleneck somewhere new, and most organisations have not yet noticed where it went.
The scarce skill is judgement, not typing
When code is expensive to write, teams optimise for writing less of it. When drafts are cheap, the constraint becomes deciding which draft is correct, whether it fits the system it is joining, and what it will cost to own for five years. That is architecture and taste, and neither is getting cheaper.
We see it in our own delivery. Time spent producing initial implementations has fallen noticeably. Time spent on domain modelling, interface design, review, and testing has not; if anything it has grown, because there is more candidate code to hold to a standard.
Generation is cheap. Verification is the job.
Three consequences worth planning for
- Test suites and type systems become more valuable, not less. They are how you verify volume you did not write by hand.
- Code review shifts from style to consequence: does this belong in the system, and who will maintain it?
- Documentation and architecture decision records matter more, because intent is the thing generated code cannot infer.
What does not change
Distributed systems remain hard. Data modelling remains the decision you cannot cheaply reverse. Migrations remain slow and political. Security remains a property of the whole system rather than any component. None of these are bottlenecked on how fast a first draft appears.
It is also worth saying plainly: the organisations that will benefit most are not the ones adopting the newest tools fastest. They are the ones with clean interfaces, real test coverage, and observable systems, because those are the conditions under which you can safely accept a large volume of change. Teams without them will find that faster generation mostly produces faster mess.
How we are preparing
By investing in the parts that are becoming more scarce, not less: stronger typing, evaluation harnesses, better observability, and a review culture that argues about design rather than formatting. The rest of the tooling will keep changing every eighteen months, as it always has, and that is fine.
About the author
Angel Maile
Angel co-founded Bonang Technologies after a decade spent building software inside organisations where the technology decisions and the commercial ones were made in separate rooms. The company exists to close that gap: engineering that starts from what the business is actually trying to achieve, and leadership that can hold both conversations at once.
