Jeremy McGee
David VujicIs there a "Modern Waterfall" rising in the new landscape of software development, or am I misunderstanding?
I wrote a post about it:
https://davidvujic.blogspot.com/2026/03/agile-agentic-engineering.html
ben@davidvujic much of “AI” engineering seems to involve throwing away decades of knowledge and starting from scratch, so it’s no surprise if LLM users have forgotten/ignored all we’ve learned about waterfall versus agile too.
(Comparable to the security problems too, which seem to me to be replicating problems from the 80s and 90s that we’ve known how to avoid for decades.)
@benjamineskola I agree, and wasn't it someone at one of the AI tooling providers that said something about we are all now junior engineers again (meaning what we know is not worth much anymore)?
I see it as our experience and knowledge is a huge advantage. The trick is to keep an open mind and keep on learning.
@davidvujic My feeling is that what we (as an industry) know it worth a lot, but the AI advocates don’t value that knowledge.
In fact it’s not even new to AI advocacy but Silicon Valley startup culture more broadly has the same problem. The old knowledge is irrelevant, they say — and then encounter all the same problems over again.
In particular I’d assume any claim made by the AI companies themselves is marketing. These tools are *so* powerful and *so* different that everything has changed — or so they want you to believe.
OldProgrammer@davidvujic Interesting. I think that big planning is "coming back", or makes more sense now, since the speed of execution has gone up in recent years. And with agentic coding, planning is a need; the agents do not know what to make without a plan.
At my work, we do scrum, but looking at the bigger picture, the product/UX, marketing, and sales teams have done a month or two of planning before the development team starts working. Then we do two weeks of work, then demo and retro. Adjustment happens during development, and you could argue that's what makes it agile. But we still have a big planning phase. It all looks a bit like a waterfall of smaller waterfalls.
I don't think big planning is an issue. I think the execution matters more. If management throws two months' worth of work into a pile and then lets the developers fight it out, bad things happen. Going agile when executing will make big planning a success.
@hugo I have the opposite experience and hope big planning isn't making its big comeback 😁
@davidvujic Maybe it will be a different kind of big planning coming. I see two kinds of planning:
1) Planning of how to execute - that's the bad kind of planning when done on a big scale. The waterfall kind.
2) Planning what to execute - that's the kind of big planning I see coming.
Maybe some will be doing both or a strange mix of both into some monster plan 👹 that is doomed to fail 🙀
I once heard a Navy Admiral call number 2 for the "In command, but out of control" kind of operation. You leave the execution up to the experts.
Orchun Kolcu@davidvujic
If the coding part could be easily repeated (i.e. infinite tokens) and the LLM is producing code at least 6x, 7x the speed of a human, why not?
Caveats: no real way to control the low-level design of the code - you may end up with something that's only maintainable by LLMs, but nobody seems to worry about that.
Caveats: no real way to ensure security or correctness, both bugs and security holes are to be expected, but nobody seems to worry about that either.
Of course at some point "infinite tokens" will clash with reality of the inference cost.
@orchun What I'm thinking about is that the hard part is the big upfront planning, as it always has been. That's why most software teams have abandoned that workflow since long time ago.
If we produce code at 7x speed, but the wrong things, we'll end up in 7x waste. Agile principles should work very well even if there's high perfomant agents involved.
This is basically what I write about in the post.
@davidvujic
What I meant was they don't care about the waste, because right now it's cheap to throw it away, change the spec and regenerate.
What I meant was they don't care about the waste, because right now it's cheap to throw it away, change the spec and regenerate.
@orchun Aha, I see what you mean. Thank you for clarifying!