been using AI coding agents for a while now and honestly the workflow shift is wild
like 6 months ago i was still writing every line myself. now i basically describe what i want, review the diff, and steer it if it goes off track. not autocomplete. actual agents that read the codebase, edit multiple files, run tests, and fix their own errors
the weird part is my job is becoming more about reviewing than writing. i feel like a senior dev managing junior engineers who never sleep but will absolutely do something stupid if u stop watching for 5 min
shipped a feature last week where the agent wrote the backend, wrote the tests, caught its own bug, fixed it, and opened the PR. i just reviewed and merged. that hit different
what's ur stack looking like these days. cursor? claude code? terminal agents? genuinely curious what people are actually using day to day vs what's just hype
Top comments (0)