I noticed a lot more joy using AI from people at smaller companies or working by themselves :)
I say this as someone self employed that burned almost $1000 on tokens last month. And had. A lot of fun doing it.
back
6 comments
No surprise. People like being more productive when they reap some of the benefits of that increased productivity. If you're expected to be 10x more productive but don't get a raise, all you're doing is stuffing money in some executive's pocket while your job security goes down.
I'm being heavily consulted to advise management on culture change towards AI. And my number one message is this: make the number one, first and potentially only beneficiary of AI use the individual staff members themselves. If they have more time now, DO NOT start filling that with more work for them to do. If they do more all by themselves accept it as a bonus (experience says this is overwhelming what will happen anyway). Whichever way it goes, let them experience directly the benefit, and let the culture change happen organically downstream from that.
I think all these companies front-loading staff reductions are actively sabotaging themselves in the worst possible way in this regard.
I would love to hear more about your advice and the coaching you are giving to management. We also have a strong push to prove evidence of climbing productivity with clearly state future staffing goals. I would like to advocate for this, at even partially, enhancement and quality of life improvement for IC folks.
It starts with the generic pitch around culture change - "culture eats strategy for breakfast" style. Then a bit of shock and awe around how extensively AI is going to redesign business processes in the long run, leading into an argument about it being a marathon, not a sprint and at the moment everyone is treating it like a sprint, the real winners will be those gearing up for endurance. Then structuring the pathway: personal productivity as a cornerstone ebbing into pilots of implementation in areas highly aligned with AI capabilities minimised risk - all as preparation for the main game which will ultimately redesign core business processes in an AI first way.
I will say I am a bit of an outlier. I see others mostly pitching for things like small teams of "AI Champions" etc. I don't favor this because I think it will lead to dysfunctional outcomes (people trying to make the initiatives fail because they weren't "chosen" etc). So I pitch for the broad based, whole organization journey etc. But it does require a strong argument for acceptance of a slower pace of externally visible adoption.
This.
I’m in a dreadful situation right now. Everyone in team got a claude account, but I’m a contractor so not for me (the only dev in team of 25 consultants). Someone in the team assigned me a task to review claude skill that opens up tickets for me. I’m not even using claude and official policy is no AI use for development…
Otherwise it’s been mixed bag. Pace definitely picked up and things that I actually enjoyed doing (UI) it does very well. Things that are actually hard (backend logic) it sucks and painted me in corner too many times.
Meta is on the extreme other end of this. The article opens with how they're now using AI to monitor how everyone uses their computers.
It's still insane to me that Meta thought this would be a good idea, or that employees would be comfortable with it even though they claim it's only used for anonymous AI training.
> using AI to monitor how everyone uses their computers
It's the other way around -- they're monitoring the computers to train AI.
probably both, to be fair.
Meta may know that their employees will put up with it, given how depressing the job market is right now, but unhappy, cynical, resentful employees do not produce good software and innovations.
there's a real financial cost to treating devs like cage-raised livestock.
It's unclear how you would use LLMs to monitor clicks. Unless you just mean they're authoring the monitoring software with LLM assistance (which is probably right).
LLM generates context based on what's on the screen and associates it with the action taken by the user. It is less "point of time" but more "charting the flow"
For example. page content of a PR with open comments, next action is to focus on the first comment. when a new PR with no open comments is shown the approve/push button is the next action. That starts a re-enforcement loop.
Could this be a vector to poison the AI? I am not one for sabotage, just bad karma all in all, but not all are like that, and if one knows their days at ACME are numbered, the sirens start singing.
If they were competently evil they would have just done it quietly.
I work by myself and feel no joy in using AI.
I work by myself an feel great joy. Today I talked to the AI about a feature I want to add to this week's project (https://www.writelucid.cc) and it had some good feedback. Later I refactored a big part of the code to simplify it (though I had to explain to Claude why this was possible), and it came out great.
I've never been happier, I can now build everything I've been wanting to build, really fast, with very few bugs.
I work for myself and I absolutely love AI.
I'm able to get 3x the work done. Greenfield stuff appears almost immediately.
My job is providing value to customers, not worshipping at the cathedral of software that will last forever. Nothing lasts forever.
Start treating software as ephemeral. It'll click.
This doesn't mean write low quality, unmaintainable software. It just means focus on getting stuff to your customer.
Writing in super typesafe languages with the highest level of strictness helps a lot. My AI stack is Rust and Typescript.
I tried using it last week to make a simple Yu-Gi-Oh! website, that shows decks, lets you rate them, register users, etc. kinda like masterduelmeta.com and I enjoyed using it, but definitely did not enjoyed making it. I didn't felt a sense of ownership or dopamine from nailing the styles just right, or making the cards shimmer when you hover them.
All jobs can generate income. What led me follow this one job in particular was the joy of turning nothing into something, and it now feels that the most effective way to do that is for only $99.99/month, and that price needle is only going to move further upwards as capabilities increase.
> it now feels that the most effective way to do that is for only $99.99/month, and that price needle is only going to move further upwards as capabilities increase.
That's not how economics works.
That can happen briefly with monopolies and ossified markets, but there is typically always an alternative that will seek to break in and grab market share.
Chinese tokens are pretty cheap and they'll gladly undercut US hyperscalers.
This is the right way to look at things now. It might not always have the right track record, but AI built coding is more likely to have all the right permissions in place by default, most likely to copy existing patterns in your codebase, most likely to use the highest performance patterns and on top of all that, the spec will match what was asked of it.
What magical AI are you using? That’s not my experience at all.
I use cursor on auto mode almost all the time. I switch to Opus 4.7 when I need I know it will go off the rails. But generally Auto mode "just works".
For my personal work on my own projects, Codex 5.5 because it's cheap at $20 / month and I get in about 10 prompts during my work day (would be more like 40 if I was not focused on work though)
Claude with the 4.7 model is getting pretty good.
there is a significant learning curve to using AI well. learning to stay skeptical and keep your brain on, developing an intuition of how much free reign to give it, writing ironclad specs and design docs and keeping them updated, making work easy to inspect, the tone you use talking to it, using one agent to critique another's work, etc.
basically, AI will produce slop if left unattended. but it's not really its fault.. it's a process failing, like not supervising the interns. using AI the Right Way(tm) is a mental workout, quite a bit slower, but extremely rewarding (ime.)
I can't even get LLMs to reliably use tool calls instead of bash, let alone follow existing patterns in a codebase.
What do your prompts look like?
Mine are pretty robust and articulate. I tend to write very lengthy instructions and include snippets of code, file paths, struct names, etc.
Been feeling that energy too, trying so hard to stay at my current big co job for the health insurance. But the draw is pulling me hard.
I've generally assumed that AI would make developers get lower compensation because of the lowered quantity of developers required for the same output, but this raises the possibility of it actually increasing if more developers end up doing their own things instead of entering the broader labor market :)
It could increase compensation by growing the economy. (E.g., perhaps counterintuitively, skilled immigration has this effect.)
the problem is that very few to none SWEs “doing their own thing” will ever make a penny out if it. whatever they do, if it actually makes a little traction, will be cloned and copied in a week by someone else. this whole idea that “we’ll see a 1-person billion dollar startup” is as silly as it gets
What’s your ROI for that $1000?
Just wait until Big AI copies those businesses.