@morganlinton How much money or VC funding do you have? Because running loops means you will be paying A LOT in tokens

@morganlinton Create a background agent that reviews PRs and auto implement suggestions, asking whether the implementation matches the intent. Do this with fresh context up to N times

@morganlinton now we have to be "loop coder" 🙂

@UriGil3 @morganlinton Complicated speak for something that’s very simple. Figure out what you do regularly. Figure out how to get AI to do it. Have AI automatically do it regularly.

@morganlinton what does it even mean to write a loop. what does that look like

@morganlinton What does it mean

@UriGil3 Ask whatever agentic coding agent you use to teach you, or ask Grok to summarize what Boris and Peter have been tweeting about w/respect to moving from prompting to looping

@alexanderrX_ Haha, yeah - everyone should do so at their own risk. I only use frontier models for the last mile, I’m a multi-model guy, so any looping would be done on a cheap or free model.

@morganlinton Going from prompts to loops means you stop playing with adjectives and start building actual state machines. It's basically a transition from creative writing to systems engineering.

@morganlinton oh right lol I should have started with that

@petergyang @grok break it down for Peter

@morganlinton @rohanpaul_ai True, but opportunities as well 🫠

@mrclhnz @rohanpaul_ai Aka tokens go brrrrr

@morganlinton I just saw @steipete tweet about that. Time to go down the rabbit hole to figure that out.

@morganlinton Are we talking about loops waiting for information or loops iterating on something? Like ralphs. *haven't looked at the video

@morganlinton @rohanpaul_ai Why not just ask Claude?I’m

@morganlinton its /goal

@rohanpaul_ai @morganlinton Loop Engineering

@morganlinton The prompts kick off loops it's not really "zero prompt"

@morganlinton Actually that’s a good idea. I usually keep it on gpt5.5 xhigh for everything because I am too lazy to keep changing it.