source: simon willison: not so locked in any more
level: technical
simon willison reflects on a conversation at a recent conference with a developer from a medium-sized tech company. the company had just used coding agents to rewrite their legacy iphone and android apps into react native. when asked why they chose react native despite coding agents potentially lowering the cost of maintaining separate native apps, they explained that react native had improved enough to meet all their needs. more importantly, they felt that if the decision proved wrong, they could easily port back to native later.
this echoes a point made by mitchell hashimoto about bun's migration from zig to rust: programming languages used to mean lock-in, but that is changing. with ai-assisted tools, the cost and effort of switching languages or frameworks drops significantly. companies can experiment with different stacks without the fear of being trapped by an early choice. the ability to rewrite codebases quickly using agents reduces the risk of technical debt from language decisions.
the shift has broad implications for software development. teams can prioritize developer experience, performance, or ecosystem fit for a project, knowing they can change direction later. this flexibility could lead to more innovation and less dogma around language choices. as coding agents improve, the barrier to rewriting entire applications will continue to fall, making technology stacks more fluid than ever before.
why it matters: ai coding agents lower the cost of rewriting software, reducing the risk of language lock-in and enabling more flexible technology choices.