Raw LLM Responses

AI isn’t a binary threat-or-saviour issue for software teams. Its most reliable impact today is that it raises the productivity of the people already in the development process. Modern code‑generation tools can be used to produce slop if used naively, but they can also generate clean, standards‑compliant code when guided by strong prompts, guardrails, and well‑defined engineering conventions. In practice, AI amplifies the capability of disciplined teams rather than replacing them outright. If headcount reduction was the goal, it would probably be an ideal that AI did not generate code, as that is a human requirement, that in theory is an unnecessary step for machines. Even if an organisation adopted a fully "codeless" or AI‑assisted development pipeline, humans would still be required to specify behaviour, validate outputs, audit safety and compliance, and test the resulting system. These tasks don’t disappear when code is generated automatically; in many cases they become more important, because the system’s behaviour must be verified rather than assumed. You can already see this dynamic in domains like autonomous vehicles (AV). When an AV system mishandles an edge case, engineers don’t simply patch a single line of code, they adjust data, refine models, and retrain components. But this doesn’t eliminate conventional software engineering; it sits on top of a large, carefully maintained codebase that handles planning, mapping, safety logic, and system integration. The work shifts, but it doesn’t vanish. So the question isn’t whether AI will erase development teams. It’s whether organisations use AI to extend what their teams can deliver or to compress the scope of human roles. The technology itself doesn’t dictate the outcome - the organisational strategy does. Any CEO that is intending to use AI to reduce the development team headcount, should perhaps have a chat with the AI, it might suggest that the AI is used to make the development team more productive, and to change the CEO.