Code review is still the bottleneck because agents lack your team's shared understanding of the software. Without that, there's only so much you can trust their judgment. Trust is the real bottleneck. Giving agents that shared understanding is the next big challenge.

I don't think we realise the cognitive load skills puts on agents. A typical code review skill asks an agent to keep dozens of things in mind before it's even started reviewing anything. Well defined tasks are useful, but I am pretty sure there's a better way.

Your model solved a maths problem we've been working on for a century? Cool, I guess... Can it one shot videogames tho?

Software's peculiarity is that it's never really done. What you've learned and the decisions you have made are what make it better. You can clone a snapshot of what those discoveries produced, but you'll still struggle to catch a team that's been building something for a decade.

I could probably learn how to make my own shoes. I could probably build a machine to make my own shoes. I might be able to use an agent to get the shoes made and on my feet faster. I am almost definitely going to keep going to footlocker because I like shoes but do not want to be a shoemaker.

I can see a future where professional yak shaving becomes its own thing for engineers outside of the traditional lifecycle of software dev. People continuously honing the feedback and discovery cycles of the machine. Lots of people were already kinda doing this anyway.

Are the people putting out 2500 prs a month just doing tiny little micro-optimisation PRs all the time? Is any of the work actually good? The feature factory thing feels achieveable now, but I feel like just maintaining it becomes the job at that stage.

Me when I see one of the good accounts post with quotation marks that aren't on a keyboard.

I think people are getting a whiff of "let them eat cake" smell from the flight booking demo. Most people's air travel isn't for work, it's for holidays. Which, most people I know enjoy doing. There's value in taking something inaccessible to most people, like an executive assistant, the favoured tarpit of the decade, but democratising the solution is only useful if you're solving problems your users actually have. The second half to that post they didn't do was figuring out what regular people actually want taken off their plate. Rich people are probably keeping their EAs.

I've tried most of the popular Clawdbot-ish things and never really got enough out of them to change what I already do. Dots not clicking was surprising because I already do a lot of the stuff it is meant to do with ChatGPT and Codex. I am not sure that persistent agents are really going to give you a better experience. ChatGPT/Codex's memory has been really useful. I can talk to it and it already knows what I'm talking about, then spin up agents for whatever needs doing. None of those subagents really benefit though from having the coordinator's persistent memory. I don't need or want the plumber I hired to know what type of yogurt I like. A lot of the "AI life OS" stuff is kind of working backwards. They're trying to build bespoke agents, when what they should be building are tools that give agents a better mental model of who their user is and what they care about. I think the better end goal for OpenAI, rather than a new product, is that regular people don't notice when ChatGPT becomes their Hermes agent. It just happens naturally that they end up delegating more and get better results. I don't think most people expected to be running their businesses out of Gmail when they first signed up, but here we are.