I have started to feel more doubts about the benefits of using agentic AI for coding (like in coding harnesses like OpenCode, Pi). It seems like the next step for AI—after all, isn't more AI always better? But I have had mixed results at best.
It seems like everyone has decided that agentic coding is the future, but is this necessarily true? I agree that AI has benefits in software development. But it can be useful just in chat bot form, where the developer generates a required string manipulation function and then pastes it into their program.
Having an agent edit and compile files seems impressive, but are these things humans really need help on? I have never met a developer who cannot edit files or compile programs.
Using chat bots to generate functions, meanwhile, has some utility—it saves time. And because it is a fairly lightweight task, it can be run locally (I use gemma-4-26B_q4_0-it.gguf in llama-cpp) for free. It might be that I just do not have en expensive enough subscription, but I seem to get better results from chat bot-style development.