I've been on quite the agentic coding journey today. I was really hating the back-and-forth I was doing with my agent on refining a problem space, I made an agent to check its work before it got back to me. That only solved half the problem. I noticed something funny in the responses though—my first agent told me it went back and forth with the review agent. When I eventually got fed up with the poor quality response I was still getting, I asked to make a panel of adversarial agents arguing different opinions and coming to a consensus and returning their results. 100%? No. Ages better than a single agent? Heck yes. Adversarial agent panels are my "always start with a clean context” advice brought to life in a single long-running context. I dig it.#ai #vibecoding #webdevelopment
Related
Re-watching Frieren, I just had an epiphany: demons evolved to use human language to fool humans, all the while not unde...
Re-watching Frieren, I just had an epiphany: demons evolved to use human language to fool humans, all the while not understanding any of the underlying actual meaning of the words....
👀 Ah, the age-old quest of "Who does #Anubis actually stop?" - a riveting tale of a tech-savvy #Sherlock attempting to o...
👀 Ah, the age-old quest of "Who does #Anubis actually stop?" - a riveting tale of a tech-savvy #Sherlock attempting to outsmart an #HTTP #proxy with a patch, a kernel, and a sprink...
Running a 28.9M parameter LLM on an $8 microcontrollerhttps://github.com/slvDev/esp32-ai#ai #github #llm
Running a 28.9M parameter LLM on an $8 microcontrollerhttps://github.com/slvDev/esp32-ai#ai #github #llm