The technological shift that AI has brought us is not just about new tools. It’s a paradigm shift — and it demands a mindset shift in how we observe and build software.

Software used to be imagined as solid.
A mountain: hard, fixed rock.
Or a Lego system: piece stacked on piece, logic chained together neatly.
Deterministic. Predictable. If-this-then-that all the way down.

That world isn’t gone, but it’s no longer the whole picture.
We’ve introduced a new element: water.

AI is fluid. Non-deterministic. It fills containers, flows through cracks, resists simple control. You can’t hold it in your hand without some slipping out. You can’t always see how deep it goes, or what’s underneath. Unlike a mountain, you can’t just stand on it and feel stable. Water is alive, moving, and in constant flux.

---

the new element

Mountains will still be part of the software lifecycle. There will always be rules, gates, hooks, chains of hard logic. But alongside them we now work with water — probabilistic systems, embeddings, tensors, semantic meaning. This isn’t just three-dimensional complexity; it’s infinite-dimensional. Almost impossible to fully picture.

Treating AI like it’s still rock is a recipe for failure. You can’t chisel water. You can only redirect it, contain it, channel it.

---

holding water

When you cup your hands, water slips out. That’s what it’s like working with LLMs. They don’t fit neatly. They’re black-boxed. You can’t trace every “if this then that” path inside them. They’re statistical flows, not logical ladders.

So you need new strategies. Build the container, not the rock. Create the conditions that guide flow. Harness, don’t hammer.

---

the shift

If you try to force AI into the old paradigm, you’ll break your systems. But if you learn to hold it differently — to respect its fluidity — you’ll have the advantage.

That’s the real shift: recognizing that we now code with two elements, not one. Rock and water. Rules and flows. Solidity and fluidity.

Mastering both is the key to surviving — and thriving — in the AI era.