How can companies adopt AI coding tools in a way that is sustainable for software, engineering teams and developers? Mistakes that are happening now.

Posted by haricane8133@reddit | ExperiencedDevs | View on Reddit | 19 comments

I recently wrote a long article with my thoughts on how AI-assisted software development is currently being handled inside many companies. As a relatively junior developer, my thoughts and obervations may be wrong; posting here to pick your brains and experience. Topics discussed in the article are 1. Sustenance of software, teams, developers - how do we regulate usage of proper AI so that we do not self destroy ourselves 2. What are true innovations? Why are developers pressured to showcase every minor win? (which just might be using a claude feature) 3. There are a plethora of tools and feature? What is a good way to learn and grow in AI as a software engineer. 4. How to measure success? Why wrong / poor AI usage metrics collection by top management, are causing damage. This is not an anti-AI take. I use AI coding tools, and I’ve gone through my own small version of the agentic coding hype cycle: from refusing to use ChatGPT, to getting excited when I could leave an end-to-end SDLC-style coding workflow running overnight, to eventually realizing that even that was not enough. I strongly feel that the developers who benefit most from AI are not just people who can use AI agents well. They are people who already have strong judgment across system design, debugging, domain knowledge, legacy codebases, tooling, infrastructure, and how LLMs behave. The pressure to use more and more abstracted high level multi agent orcheestration tools or harnesses is growing, which doesn't seem to help as well. Here is the article itself - [https://x.com/haricane8133/status/2062196783330312296](https://x.com/haricane8133/status/2062196783330312296) . Would love to have a conversation, or just feel free to vent about what you are going through!

Reply to Post

19 Comments

[-]

cbusmatty@reddit

Number 1 is a useless point. The top performers who figure it out will not listen to a nameless faceless “us”, and we will build the future of infrastructure and engineering. You can be on the front of it or the back of it but you cant stop it. 2. This isn’t ai related and has existed since the dawn of development, it’s just now magnified because of how valuable and useful these tools are. 3. Again, not a new thing - you should always be learning and sharing. The easiest way to learn is to force yourself to do something outside of your comfort zone. Pick up a new tech, work with someone on a new project, volunteer to present. Now you can expand that to ai tools and problem solving literally anything. 4. Again since the beginning of developing software success measurement has been a hand wavy thumb on the scale: this is ai exposing poorly run organizations and processes. Most of the problems are organizational, either leadership or strong staff engineers, and ai just puts a magnifying glass on problems you already have

[-]

haricane8133@reddit (OP)

The problems may be already existing and AI may have magnified them to a very large extent. Valid point on big corporations not listening to "us". Do you have any thoughts on how you would solve it? Did you do through my article?

[-]

cbusmatty@reddit

I did read it but it’s missing the point - development, writing software is effectively solved regardless of what anyone says in here. And highly skilled and trained engineers can apply enterprise context and practices which you’re not going to find on Twitter or Reddit. What’s missing is everyone talks about engineering, but the real focus is on what you’re delivering. You have a bunch people working off of kanban cards instead of understanding what they are delivering to their business partners and leaders. The goal of a good engineer right now should be organizational direction, and tools, libraries, process, workflow that drive organizational success and raise the floor of the leaders iq. And the solutions ai can help us build and the research we can use will help enable them. The first person in an org going through any of this mess who solves a problem for the area or organization instead of just completing their card for the week gets included in the next conversation and the next. You will never be as good as that 14 year sr engineer at Java but you now are on a level playing field with an opportunity to catch up in value and knowledge

[-]

haricane8133@reddit (OP)

Okay thanks for explaining. What I'm hearing is, skilled engineers will figure out what to, regardless of AI, right? The article is just my thoughts for the*rest of us* mortals (who aren't that skilled engineers) to try and focus on the right areas to get to that place, instead of being washed down with AI pressure (which you mentioned in a comment back, was just regular pressures amplified with AI) Instead of just saying, "survival of the fittest", I'm trying to see if everyone has a way out

[-]

cbusmatty@reddit

Every engineer of any skill should always be working on: Networking + helping others and teaching others, understanding what your boss is judged on and their boss, and making sure your work aligns to those goals, the current news and capabilities of your technical domain which also now includes ai for everyone too, and career goal you’re working towards.

[-]

haricane8133@reddit (OP)

Good. These definitely help any person from any career climb up the corporate ladder. Any ideas on how one can grow in that "skill" you mentioned, in your "engineer of any skill"? These are two different skills we're talking about. Skill to grow career and skill to grow technically. I agree there's a sight overlap. The premise of my post was to make sure this "technical skill" is what engineers are focused on, and not drowning themselves in AI

[-]

haricane8133@reddit (OP)

Also don't you think that if we managed to get a few good pointers or path across through this paltry reddit channel or Twitter, there might be a few engineers who get the trigger and get benifitted? That includes me too! If someone here provides a solution, an advise or a way forward, it'll help me too! The article missing a point is exactly why I reached out here. To know exactly what else....

[-]

Additional_Rub_7355@reddit

Most problems are indeed organisational and I don't see how ai can help in any way so far tbh.

[-]

haricane8133@reddit (OP)

Any thoughts on how you'd solve these organizational issues? As the other comment mentioned, we may not be able to change anything. But just wondering

[-]

Additional_Rub_7355@reddit

Not even my managers know mate. Some pretend they do but that's another story.

[-]

cbusmatty@reddit

The answer everyone hates here is that your high performers will figure it out, and leverage that with high emotional is to translate it into leadership understanding and guidance. No more codemonkeys, you must think as a leader and act like a leader. The market is yearning for confident engineers who value guardrails and security but have a good path to what forward looks like avoiding pitfalls and doing more business

[-]

Additional_Rub_7355@reddit

You appear reasonable, no idea why you being downvoted.

[-]

haricane8133@reddit (OP)

The funny thing with Reddit in general (and also this sub) is that posts that just complain or vent about a problem get tons more action than posts that attempt to solve that problem 🤣 Let's work together folks!

[-]

Additional_Rub_7355@reddit

I loose cognitive capacity and knowledge to gain short-term speed when i use any form of AI approach to do my work. That's how i view it and that's my general opinion as of now. We will see...

[-]

haricane8133@reddit (OP)

Yeah I agree that any automation has the potential to do this, and AI is a huge next level... Do you have any ideas on how this problem can be solved?

[-]

Additional_Rub_7355@reddit

On an individual level? Don't allow more AI into your life than the amount dictated by your employer. There are already devs right now that forgot how to code, they will also forget how to think at some point. Don't allow yourself to become that i guess, ai is not going to solve the tough problems of our profession anyway.

[-]

haricane8133@reddit (OP)

Yes! What I wrote, is just the elaborate step by step version of how to do what you said, for different people, for juniors, seniors and leaders. I also argue that companies need to slightly ease on the AI pressure... Not sure what all the unlikes are for though...

[-]

haricane8133@reddit (OP)

No AI was used to generate the article I linked!

[-]

haricane8133@reddit (OP)

Folks stop it lol. It's not funny. I really typed out the full article. Declared it here as a comment because the sub guidelines asked me to