Ask HN: What is your #1 practical lesson or "aha" moment from coding with AI?

As a developer, my own workflow has shifted completely over the last few months, but I feel like I'm still uncovering the best patterns. I'm curious about the specific inflection points where things clicked for others. Was it a shift in how you prompt, a change in your tech stack, or learning when not to trust the output?

▲

rimeice 13 hours ago | parent | next [-]

Don’t use it for coding. Well at least using it for everything around coding rather than just coding itself was my aha moment, I realised I could understand new codebases way faster, brainstorm, sense check ideas draft better thought out arguments to my perspective and generally ask stupid questions. All of which has significantly increased my learning as a developer. Using it for coding itself is another matter.

	▲	qsera 6 hours ago \| parent [-]
		I have been saying this for a while. Using LLMs for reviews and for learning new things gives maximum long term bang for buck. If you use it for coding, that would be like killing the goose that lays golden eggs in the hopes of catching some kind of jackpot.

▲

kasey_junk 14 hours ago | parent | prev | next [-]

The chat metaphor held me back for too long. These are black box functions that produce text and should be treated as such.

It came to me gradually that I was slowly replacing each part of my skills with scripts that took defined inputs, validated them, triggered agent sessions with defined prompts and spit out validated outputs in defined formats.

Now I do it proactively, if it looks like a function or a loop, or workflow encode it as such. My outcomes are better, I can use cheaper models and it’s easier to test.

▲

kingkongjaffa 14 hours ago | parent [-]

Do you have an example?

	▲	skydhash 13 hours ago \| parent [-]
		Not GP and I’m a general AI opponent, but it is the same strategy I adopted at work due to mandates. I can’t give you a past example(they’re all related to $work). But you can probably work out a skill that downloads an emacs package, extract all user options, provide general configuration frameworks, prompt you with those and then write the use-package incantation for that. The pattern is to systemize your decision making, figure out decision node, and then provide a framework to solidify that structure. It will rigidify it, but you gain in speed that way. Basically writing like the AI is an idiot. It’s not, and it’s not intelligent either. Bit you have to make the process explicit. In most cases, a deterministic script would be better. The AI advantage is flexibility. In the above example, the flexibility is required for the configuration recommendation. Everything else before that component, and after should be very explicitly and even specified as small software to make it stricter and less prone to errors.

▲

jaredXX 14 hours ago | parent | prev | next [-]

Building a web app in a language I simple couldn't code in if put in from of Vim terminal (Flutter). The ability to iterate and express ideas at almost the speed of thought must have been what passengers felt like the first time they got in an airplane or car.

▲

dredmorbius 13 hours ago | parent | prev | next [-]

This Ask is quite similar to one which ran a couple of weeks ago, "Ask HN: What was your "oh shit" moment with GenAI?" (1,182 comments):

<https://news.ycombinator.com/item?id=48406174>

Unless HN mods want to add this as (another) regular recurring feature, it might be considered a dupe.

	▲	AnimalMuppet 11 hours ago \| parent [-]
		No, I think it's a different question. That was "what were you sure that AI couldn't do, and then you saw it do it". This is "practically, how do you make it work"?

▲

downbad_ 13 hours ago | parent | prev | next [-]

This is very similar to a recent Ask HN post that made the Frontpage.

Ask HN: What was your "oh shit" moment with GenAI?

https://news.ycombinator.com/item?id=48406174

▲

hgsocket 12 hours ago | parent | prev [-]

getting know lot of new tech that actually have been around for years