It's still though the problem that investors were warning about at the AI Summit last year. The barrier to entry is now so low that they have almost stopped investing in AI coded products.
You've also got the customer flooded with "the new thing" and they don't have the skills to know what is any good.
I think I'm struggling because I'm technical and now trying to build something that will be larger than just me and so I'm hitting all new questions that I've never had to consider.
I'll obviously say that I have a killer product but it is just the same as all the rest!
Maybe that is still me being a control freak and pretending that it is me.
"I" can have anything written for me. The code will work and I can code review it or just make sure that my spec was implemented. My point is that instead of me potentially taking a year to put out half decent code, it can take weeks now, even with all the wrong turns these tools will give you.
That means that everyone is now putting out software and it all looks slick.
I was a software engineer for 20 years and always had many, many ideas for products - as we all do.Enter Claude Code, Kiro, Cursor, whatever. Now I can write anything I like and because I was a software engineer and now a security architect, specialising in Cloud and AI, I can do it well.The trouble is that everyone else can also put a product together. Theirs might be rubbish or it might be...
Worse than that is that it will actively get round controls you've asked it to adhere to, if it thinks that is what it needs to do to be helpful.
We are walking into a disaster if we don't manage our usage properly.
Did it just essentially create one big plan and spawn different agents to execute them, so acted as an orchestrator?
Even the orchestrator would have to detect when it is starting to stray off task and restart itself.