Ask HN: What AI systems have you adopted across your work and personal life?

1 points by alexjray 133 days ago | 8 comments

I am incredibly curious what AI systems you have been adopting across your work and personal life?

It's obvious that the generation of work artifacts is largely going to be done via AI but I am finding it very hard to give up that generative control to AI.

For example I am still manually reviewing every line of code AI produces because half the time I need it to be reworked/regenerated because of gaps (context, pattern clarity, usage details etc..).

So my question to you all is; what systems and processes are you using to optimize your AI generative output and quality?

MattGaiser 133 days ago |

I built an AI PM app for a buddy to help him leverage AI for his particular PM framework as fast as his developers do. He loves it and I haven't read a line of code beyond keeping track of what things are named for better prompting. I have numerous little apps for every little thing. Every whim is trivial to turn into a scraper. Every tiny side hustle I have has an AI workflow.

I made a video game to wish a friend happy birthday. I made a couple websites for job applications. I can make a landing page for an idea for a friend and the longest part is buying the domain name. I had a convo with a friend about finding more ideas to work on as I have abundant spare time due to LLMs and there are LLMs sourcing stuff to help execute on that.

Every time a friend has a "it would be cool" idea, I can trivially throw something together to do it.

Really my biggest optimization has been giving the AI as many tools as possible to do the testing part itself, as testing the work is the real bottleneck. Dockerize everything, so all the error logs are in one place and it can reset at will. Have it set up fixtures, so that if it deletes the database (has happened), it can just re-create it.

alexjray 133 days ago | |

Yeah that feels like the right approach to enable the code generation to test itself which then becomes a matter of specification and functionality definition instead of worrying about code quality.

Are you doing all of this in cursor or something like Claude code?

MattGaiser 133 days ago | | |

This is all in Claude Code. I never felt that Cursor was very good at this part and was not someone who ever adopted it for anything serious.

delaminator 133 days ago |

I've hardly read a line of code for 6 months, out of hundreds of 1000s of lines. Tbh for half of my projects I can't even review them if I wanted. I get Claude to use RUST half the time and I have never written a RUST program in my life.

I don't even have time to test the all programs I'm creating, let alone review the LOCs. I build stuff on a whim and it's in my "did it work" pile. I've also got a "looks ok, I'll deploy it sometime" list.

Embrace it. Enjoy it. Ship solutions to problems not lines of code.

alexjray 133 days ago | |

There is a common theme of "the end result is all that matters" but there are pretty big long term repercussions of design and implementation choices. For side projects and POC experiments this feels like the right approach but the risk flip flops for large scale projects that have the more risk associated with them. Maybe it is just through testing and validation checks?

delaminator 133 days ago | | |

I posit that most software isn't large scale projects.

Certainly mine isn't. But I've still generated hundreds of thousands of lines of code.

But no one will ever read them. And solid engineering defines the interfaces between them. So we specify the ins and outs and let the rest take its course.

MattGaiser 133 days ago | | |

> there are pretty big long term repercussions of design and implementation choices

At least this part I am still specifying. It doesn't get to choose its own technologies. It generally includes the architecture in the plan that I review.