I once crafted a multi-agent workflow for writing code and it took 30 min to do a small task and blew my token limits, versus just 3 min if i prompted it myself. Sometimes i feel like wee fall into this hype trap of agentic coding that we overengineer because we want to see this agent run on its own complete tasks.
I once crafted a multi-agent workflow for writing code and it took 30 min to do a small task and blew my token limits, versus just 3 min if i prompted it myself. Sometimes i feel like wee fall into this hype trap of agentic coding that we overengineer because we want to see this agent run on its own complete tasks.