Trying
Started actively using AI agents right in my daily workflow. Tried three tools, each with its own personality:
- Continue — an IDE plugin, fits nicely into your usual editor, lets you plug in different models
- Cursor — a VS Code fork with an agent inside, impressive in how it understands the project context as a whole
- Claude Code (CLI) — an agent in the terminal, handy for tasks where you need to work with the file system and commands, not just autocomplete code
The feeling is that this isn't just autocomplete on steroids — the agent genuinely takes a task and drives it to a result. It changes how you think about task decomposition: part of the work you now express in words rather than code.
Still in trial — I keep probing where it helps and where it only feels like it helps.