Waterfall is quietly making a comeback, and AI is the reason
By Kris Van Hauwermeiren ·
AI makes changing direction cheaper and clear plans more valuable. That does not mean abandoning agile, but it does mean giving planning a fresh look.
For about twenty years, the debate in software was mostly settled. Agile won, and waterfall was the thing we laughed about. Big plans written up front, long phases, a finished product delivered at the end, often the wrong one. Agile replaced it with small steps, fast feedback, and the freedom to change direction. If you suggested waterfall in a planning meeting, people assumed you were joking.
So let me say something that might sound strange. AI is quietly bringing parts of waterfall back. Not the bad parts, but the core idea of thinking hard and planning fully before you build. I do not think waterfall has won, and I am not telling you to throw away agile. But the reasons we rejected waterfall are weaker than they were, and that is worth talking about honestly.
Why agile beat waterfall in the first place
To see what changed, you have to remember why agile won. It came down to one simple truth: change was expensive. Writing software by hand was slow. If you spent three months building the wrong thing, you had wasted three months, and fixing it meant more slow, costly work. So agile made perfect sense. Build a little, show it, learn, and adjust, so you never go too far in the wrong direction before someone notices.
Agile was, at heart, a smart answer to a world where mistakes were slow and painful to undo. The whole method is built around reducing the cost of being wrong. Keep that idea in mind, because it is exactly the thing AI changes.
What AI changes about the cost of being wrong
Here is the shift. When an AI can rebuild a whole feature in an afternoon, the cost of changing direction drops hard. Going the wrong way for a while hurts far less, because turning around is fast and cheap now, not slow and expensive. One of agile's main reasons to exist, protecting you from costly mistakes, gets quietly weaker.
At the same time, AI rewards something that looks a lot like waterfall. These tools build well when you give them a clear, complete description of what you want. A vague request produces a confident mess. A precise, well thought out specification produces something close to right. In other words, the up front thinking that agile taught us to avoid, because it took too long and often turned out wrong, is exactly the thinking an AI needs to do good work. Planning before building is becoming valuable again.
A second look at planning heavy approaches
Put those two things together and you get an interesting picture. The cost of a full, considered plan has dropped, because AI helps you write and test it faster. And the value of that plan has risen, because a clear specification is what makes AI build the right thing. That is a real change from the world agile was designed for.
This does not bring back the ugly side of waterfall: the rigid phases, the year long projects, the refusal to listen to feedback. Nobody wants that, and nothing here argues for it. But the healthy core of waterfall, deciding clearly what you are going to build before you build it, deserves a fresh look. For some kinds of work, especially where the goal is well understood, thinking it all through first and then letting AI build it fast may simply beat discovering it piece by piece.
So which should you choose now?
The honest answer is that the old labels matter less than they used to. Instead of picking a camp, it helps to ask a few plain questions about the work in front of you.
- How well do you understand the goal? If you know clearly what you want, lean toward planning it fully up front, then letting AI build it quickly. If you are still exploring, keep it agile and learn as you go.
- How cheap is it to change later? With AI, this is cheaper than before, which makes a bold plan less risky than it once was.
- Where is your real risk? It is no longer slow building. It is building the wrong thing well, fast. That risk is reduced by clear thinking up front, which is the waterfall instinct.
The practical takeaway is to stop treating agile as the automatic answer. For exploratory work, iterate. For well understood work, do the thinking up front, write a clear and complete brief, and let AI do the building at speed. Often the best approach now is a mix: plan more fully than agile taught you to, but stay ready to adjust the way agile taught you to.
The point worth keeping
Agile won because change used to be expensive. AI is making change cheap, and clear plans valuable, which quietly removes the ground agile stood on and hands some of it back to the planning heavy approach we thought we had retired. That does not mean waterfall is back in charge. It means the honest answer to which method is right has become more open than it has been in twenty years. The smart move is not to defend a camp, but to match the approach to the work, and to let AI do what it does best once you have thought clearly about what you actually want. Helping companies think that through, and then build it well, is exactly what we do at Binary Rabbits.
About the author
Kris Van Hauwermeiren is the founder of Binary Rabbits and a Tech Director with 25+ years of enterprise experience in digital transformation, AI and technical leadership.