Learning the autonomous coding, there are so many different skills, tools and ways and only some of them seem to work.
That means I have to:
- build something so I can evaluate the results.
- track each of these projects separately otherwise they turn into dust after quite some time. Gladly claudesidian seems to be working well with the unstructured stream of inputs. Feel like hooking it up with some task tracker cli and calendar and notifications could make life a bit better too.
- plan next projects to keep evaluating other skills and tools
It’s been discussed so many times the amount of new or personalized software that appears and will appear and it seems so true.
Whatever I built I am actively using myself - a text rewriter that cleans some of the AI speak and has MCP and cli (at https://www.refineo.app). Math teaching and solving extension at https://math.photos and a self hosted stock opportunity discovery tool that runs locally. This is just to automate what I did before manually and scale it up a bit.
> Any new ideas
There’s no product yet to cover the needs of all of us launching the software into the internet void. Any ad platform out there is a hot and very outdated mess and I just can’t.
There is going to be a better way with all the capabilities we have and someone is going to really nail it.
Thank you HN and HN community for sharing the most interesting places of internet, ideas, discussions and points of views.
Learned a lot and hope to learn more.
Don't stress about it. The junk was created the moment you bought it. Take it to the dump, they usually have some kind of electronics recycling so maybe less than 100% goes to the landfill ultimately.
Same thought came when I was reading the article and glad I am not alone.
Anecdotally, most common productivity boost is coming from cutting down weird slow steps in processes. Write an automation script, campaign previewer for marketing, etc etc.
Coding seems to transform to be a more efficient (again anecdotally) but not entirely faster. You can do a better work on a new feature in the same or slightly smaller time.
Idle time at 4% was interesting. I think this number goes higher the more you use a specific tool and adjust your workflow to that
Those are all definitely not asking about writing little services in postgres on their job apps. That tier job posting today looks more like a post for a tenure track ML faculty position.
I’ve worked at half of these places and they’re all doing yaml massaging and grpc sculpting. Not much more than basic KTLO with some regularly scheduled feature work and reliability heroics.
Well that is not what that tier job description is describing today at least. Maybe in the past or even still today if you got in early. Hard to imagine so much weight being put on ML knowledge and skills in the job app if it isn’t getting used in the role at all.
Hilarious
reply