"The reason the rich were so rich, Vimes reasoned, was because They managed to spend less money" - Sam Vimes , from Men at Arms by Terry Pratchett. The theory goes that richer people can afford a pair of boots that last longer before repair or replacement than poor people, who can only afford cheaper and less hardy boots requiring replacement in just a ⅓ of the time. It’s the old "buy cheap buy twice" idea and it still applies today, especially in the world of AI agents and LLMs. Currently your agent is 'managed' or 'directed' at least in part by a harness. In fact, the IDE coding agent is really a harness + an LLM. The LLM “writes the code” and requests tool use, the harness enables those tool calls. E.g.: Write a file or run a command line tool etc. In one sense the LLM does the productive side of the agents work while the harness has a more deterministic & feedback role. (e.g. reports back that the command the LLM suggested returned resu...
Do you know what a fall guy is? the answer is you. Why? Because when you're told that there is a human in the loop means... there is a you shaped person to blame. So, while you are working harder than ever, using the latest tools money can buy or tokens can build - someone has decided that you are responsible for all the downstream failures. It's as if shipping more, responding to customer feedback, product owner judgement calls, SRE reports from production at an ever increasing rate is not enough - they want you to review it all, and be confident (as the engineer/manager in the loop) that its "all good". If it isn't good, then you will likely be the human in the noose . The Fall Guy, A classic 1980s TV show: IMDB Any discussion on how much we should delegate to AI tools degenerates to this well-meaning and slightly nervous statement: we should keep a human in the loop. We are trained Apes, yet we are meant to hang around waiting to be blamed. The orthodox posit...