Growing, too much work to do. Every year the list of projects with really solid ROI gets longer.
AI can't troubleshoot technical issues with private company data. AI can't run a project. AI can't negotiate scope of features. AI can't make decisions on what work to do. AI can't architect system integrations and talk to all the teams involved.
AI is good at pumping out text when you give it good instructions and sometimes its even valid output, sometimes its not.
My teams work on a mobile app with 20+million lines and tens of thousands non generated lines being added every day. That is not an exaggeration. Cursor has no idea what to even do at that scale.
But even if cursor could understand the app, my other team works on the build pipelines which is another several million LOC projects that are interacting with dozens of services and used by several teams.
Maybe in a few years but currently it is completely untenable for LLM’s to work on these codebases in any meaningful way. And these are not unique, I’ve worked at many fortune 20 companies where the key product is 10 of millions of LOC of mess.
Even for personal projects cursor wrongly edits simple things like the wrong config.yaml, as it is in a different directory. I like it but it has major gaps.
I work for a medium-to-large company in the US. I'm not sure how big the total IT department is, but the software development team is about a 10-15 people, most coders.
As a senior developer, I have yet to find a way for AI to make my job as a coder easier or more productive. It has about the same productivity gains (for me alone) as a good StackOverflow search in that AI can answer questions or write some example code for me, but it doesn't go much beyond that. My employer has purchased some AI tools and if there were a product that could help me, I believe they would invest in it.
How much does AI do for you? Can it implement a feature? Mine can write simple functions but most require some tweaking and aren't particularly difficult for me to do on my own.
As an idea-guy armed with AI, what will the senior-principal do that you and the AI cannot?
I think we will see more startups with smaller teams
But I don't think its really "AI" that's the reason. It has more to do with funding availability and it being more appealing than ever to actually bootstrap and find product market fit prior to scaling up with VC money. It's a great environment to run lean and solve real problems. Not become giant and not really be sure of what problem you're solving.
There is a change in trend of-course. For example, before AI we used to hire freelance devs to build small, one time projects, or build/maintain client libraries on top of our REST APIs. But now, we can just do it ourselves! So, The impact is true.
Growing, too much work to do. Every year the list of projects with really solid ROI gets longer.
AI can't troubleshoot technical issues with private company data. AI can't run a project. AI can't negotiate scope of features. AI can't make decisions on what work to do. AI can't architect system integrations and talk to all the teams involved.
AI is good at pumping out text when you give it good instructions and sometimes its even valid output, sometimes its not.
My teams work on a mobile app with 20+million lines and tens of thousands non generated lines being added every day. That is not an exaggeration. Cursor has no idea what to even do at that scale.
But even if cursor could understand the app, my other team works on the build pipelines which is another several million LOC projects that are interacting with dozens of services and used by several teams.
Maybe in a few years but currently it is completely untenable for LLM’s to work on these codebases in any meaningful way. And these are not unique, I’ve worked at many fortune 20 companies where the key product is 10 of millions of LOC of mess.
Even for personal projects cursor wrongly edits simple things like the wrong config.yaml, as it is in a different directory. I like it but it has major gaps.
I work for a medium-to-large company in the US. I'm not sure how big the total IT department is, but the software development team is about a 10-15 people, most coders.
As a senior developer, I have yet to find a way for AI to make my job as a coder easier or more productive. It has about the same productivity gains (for me alone) as a good StackOverflow search in that AI can answer questions or write some example code for me, but it doesn't go much beyond that. My employer has purchased some AI tools and if there were a product that could help me, I believe they would invest in it.
How much does AI do for you? Can it implement a feature? Mine can write simple functions but most require some tweaking and aren't particularly difficult for me to do on my own.
As an idea-guy armed with AI, what will the senior-principal do that you and the AI cannot?
May I know which AI tools you tried? Have u tried Cursor?
Yes, it can implement features. It does 80% of the work, I still need to review and do the last 20%.
Senior-Principal guys understand architecture, design patterns, and connecting the "dots" together with business logic
I think we will see more startups with smaller teams
But I don't think its really "AI" that's the reason. It has more to do with funding availability and it being more appealing than ever to actually bootstrap and find product market fit prior to scaling up with VC money. It's a great environment to run lean and solve real problems. Not become giant and not really be sure of what problem you're solving.
There is a change in trend of-course. For example, before AI we used to hire freelance devs to build small, one time projects, or build/maintain client libraries on top of our REST APIs. But now, we can just do it ourselves! So, The impact is true.
Same as before. Yeah, I can get a lot done by myself with AI. The problem is that I don't work by myself at work.
[dead]