Finally in some ways agentic workflows magnify the power of the individual who is adept at harnessing them, they don’t have to argue (much) with the agents to effect their ideas. I’ve found a lot of very bright engineers spend their days fighting to be heard by managers and peers who can’t / won’t understand them. By unshackling them from trying to debate down idiots, they deliver way way more, and of the right things, than they otherwise could have.

This is an interesting skill plugin for me because I actually face this inverse problem a fair amount where you want to teach people about a repo and the skills associated with it so they understand the intent behind things quickly. Seeing a bunch of skill commands and behaviors doesn’t always make clear why things are the way they are. The people on the other end need context, and the rapidity with which you can create fairly complex stuff means you need a faster way than “three months of onboarding” to get people up to speed.

I would note that I don’t think I would lean towards Anthropic (operating with one 9 currently) over aws (operating with an implied 5 9’s within a region and arbitrary 9’s by composing across regions). Anthropic makes good models. They’ve yet to prove they make good operations.

I’m surprised what made you quit reading wasn’t the Claude voice sneaking through their half success attempt voice clone.

As someone 35 years into my career I agree this is the most exciting part of my career. I love programming and I do it all the time but I do it by reading code and course correction and explaining how to think about the problems and herding cats - just like working with a team of 100 engineers. But the engineers I’m working with now by and large listen, don’t snipe me on perf reviews, aren’t hallucinating intent based on hallway conversations with someone else, etc. This team of AI engineers I have can explain to me their work, mistakes, drift, etc without ego and it’s if not always 100% correct it’s at least not maliciously so. It understands me no matter how complex the domain I reach into, in fact it understands the domain better than I do, so instead of spending a few months convincing people with little knowledge or experience that X is a good idea, I can actually discuss X and explore if it’s a good idea or not and make a better informed decision. I’ve learned more in these discussions than I’ve learned in decades of convincing overly egoistic juniors and managers to listen to me about something I’m an industry authority on.

However I see very clearly we will need very few of the team of 100 human engineers I can leave behind in my work. Some of will be there in a decade, but maybe less than 1:10. This is going to be a more brutal time than the Dotcom bust for CS grads, and I don’t think it will ever improve. Mostly because we simply won’t need the “my parents told me this makes money” people, just the passionate folks remain. But even then, we face a situation where the value of any software developed is very low because so much software is being developed. It’s going to turn into YouTube where software that is paid for is very small relative to the quantity of software developed. We already see this in the last few months with the rate of GitHub projects created. If the value of any software created is low, the compensation of the creator will be low unless they’re very rare talents.

The claim you should know everything about everything you work on is an intensely naive one. If you’ve worked on a team of more than one there’s a lot of stuff you don’t totally grok. If you work in an old code base there’s almost every bit of it that’s unfamiliar. If you work in a massive monorepo built over decades, you’re lucky if you even understand the parts everyone considers you an expert in it.

I often get the impression folks making these claims are either very junior themselves or work basically alone or on some project for 20 years. No one who works in a team or larger org can claim they know everything in their code base. No one doing agentic programming can either. But I can at least ask the agent a question and it will be able to answer it. And after reading other people’s code for most of my adult life, I absolutely can read the LLMs. The fact a machine wrote crappy code vs a human bothers me not in the least, and at least the machine will take my feedback and act on it.