cmar00

Wow this feels next level! So you only talk to the orchestrator, then it calls sub agents to do the actual work, review and realign. Got a few questions for you:

- Which tools did you use to create this ecosystem?

- Do you write the tests yourself? Do you tell it which tests to write? Or does it guess them on its own based on his understanding of your preferences? How do you make sure the generated code is secure?

- As you give orders to the orchestrator and as the orchestrator gives orders to sub agents your original message might get distorted (similar to when you print a copy of a copy of a copy with your printer). Do you ever find subagents going off topic in an endless code/review/code/review loop?

I was like this too. I found pride for myself on only using the web chat. But being constantly bombarded by titles like "Look what crazy software project this accountant built with X agent!!!!" makes me worry that I'll stay behind if I don't adapt to this new model. I think after all, AI is here to stay, and it's evolving a lot faster than we can keep up

Yep, some time I think we are going back to a waterfall methodology

Thank you for your suggestion. Going straight to the next prompt while the agent is still working will probably help me stay in flow state longer. However I'll still have to context switch when checking the code. Nonetheless, I'll experiment with this, thanks!

Very good point!

So do you still use agentic programming? What percentage of code do you write yourself and what percentage do you hand off to AI?

Thank you so much for suggesting this, I will definitely enjoy reading it!

Yep, I think right now one of the best use cases is one-off disposable code for things that I don't really care about and don't want to do. At least, that's what I'm mostly using them for

Well, there's mass hysteria at the moment. I'm lucky enough to be self employed right now, but there definitely are people being forced to use them by their bosses

Yes! Yes! Yes! The band-aid patches! I really hate that.

Regarding old API regression, I think that's because the LLM sees a lot more examples of deprecated APIs throughout the code, than new replacements you just introduced. You could try forcing it to immediately replace old calls to deprecated endpoints, then actually removing the definition completely, so that you reduce the number of bad examples and increase that of good examples

I do read the commits and really try to understand what the model is trying to do. However, I think this conflicts against what we are really trying to do here: - We want to code faster so we use LLMs - We want to still be in charge, so we spend a lot of time reading and understanding every single commit and review PRs.

Before, we used to build a mental model of the code with every new line of code we wrote. That took time, but the result was a very solid understanding. Now we try to build the mental model by just looking at what the LLM produces. Similar to how students who spend most of their time practicing what they are studying with actual exercises achieve higher grades than those who just read the book, we are now at a disadvantage against our old selves who made the effort of writing the code.

Moreover, an LLM produces code a lot faster than we can read and understand. If we want to achieve an actual speedup in code output, there's no way we can read and understand everything and be faster than before. We have to sacrifice the understanding of some parts. This compounds over time and we quickly get into a place of total reliance on LLMs.

Regarding addiction: I also think these devices are addictive. They do open a lot of new doors for us builders. But I think hour grey matter hardware is not built of this speed. Give it time, the more project ideas you start spinning up, the sooner you'll reach burnout.

Hi all, first post for me here on HN!

In the past few days I found out an interesting way to do a deck shuffle in O(1) space and time complexity. I think some of you might enjoy.

Let me know what you think :)