There's no bug in my code, the bug is in the Intel architecture ๐,
And if at all there's any (which they certainly aren't), I only debug is production.
Let it go live!! ๐บ๐ป
@EOEboh I think the middle still exists. People don't just wake up having a senior level experience. They still have to go through the transitioning phase (that's the mid level) even with AI
Watched a team ship an entire feature with AI agents in 3 days.
Took 3 weeks to figure out why it was silently corrupting user data.
Nobody could read the code. Nobody owned it. Everyone just trusted the output.
Speed without comprehension isn't engineering. It's gambling.
curl/requests/axios have collectively billions of downloads a week.
Most of the internet runs on unpaid maintainer time. Starring a repo takes 2 seconds.
Opening an issue with a reproduction case takes 20 minutes.
The maintainers need the second thing more.
Stop designing APIs like internal tools.
External consumers will break your assumptions immediately.
If your API needs โtribal knowledgeโ to use, it will fail at scale.
We learned this after onboarding 3 clients and rewriting the same endpoint twice.
What's the one assumption your API made that external consumers immediately broke?
Thereโs also a weird paradox now:
AI lowers the cost of building products while simultaneously increasing the operational cost of being a power user.
Ngl, ai agents now cost more than hiring an actual developer if you're a heavy user.
There's also so many choices to pick from now and it's all enticing.