AI Can Teach You Korean. But Can It Make You Show Up Tomorrow?

Last night, I watched my child do Duolingo Korean like it was a game they had to finish.

Not “study”. Not “revise”. Finish.

And that’s when it clicked for me. The moat isn’t “content.” It’s not even “language learning.”

It’s stickiness.

Gamification sounds like a silly buzzword until you see it in the wild. Streaks, XP, little celebratory sounds, tiny dopamine rewards. It quietly turns “I should learn” into “I can’t break the chain.” Progress becomes visible. Quitting becomes emotionally expensive. That is a real product feature.

Then I hear the argument everywhere now: “Why use apps when GenAI can teach you anything?”

Fair. OpenAI and friends can explain grammar, generate conversations, correct your sentences, and roleplay a Korean café scene with infinite patience. That part isn’t up for debate.

But here’s what people skip. GenAI is a moving platform.

Today’s experience might be tomorrow’s new version, new limits, new interface, or different memory behaviour. If your learning system depends on something that changes without your consent, you are not just outsourcing knowledge. You are outsourcing continuity.

And there’s another issue that’s more uncomfortable.

GenAI is amazing at answering questions. It is not naturally built to nag you, reward you, and drag you back when your motivation disappears. It gives you help. It doesn’t always give you a habit.

Try it for a few weeks. The AI will be brilliant. You might not be consistent.

Which brings me to my favourite analogy. Cooking.

Everyone can cook. The constraint isn’t ability. It’s motivation, energy, and friction.

Cooking is not one task. It’s a whole chain:

• Decide what to eat

• Buy ingredients

• Prep

• Cook

• Eat

• Clean up

That last part matters. The cleanup is the hidden tax.

A restaurant isn’t only selling food. It’s selling relief from the full-stack burden. You pay a premium not because you can’t cook, but because you don’t want to run the entire production line and then wash the dishes after.

Now translate that back to learning.

A gamified app is like a meal kit with clear steps and a sense of completion. It reduces decision fatigue. It gives structure. It makes the next action obvious.

GenAI is like having a private chef you can summon anytime. Incredible. But the chef won’t force you to cook daily. And if you don’t show up, nothing compounds.

This is where the investing lens gets interesting.

Markets love narratives. “AI will replace apps.” “AI will commoditise content.” Maybe.

But first principles ask a more boring, more profitable question:

Why do people pay for products and services in the first place?

• To get an outcome

• To stay consistent

• To reduce friction

• To avoid the cleanup

• To feel like the kind of person who is “on track”

Most products do not win because they are the smartest. They win because they make customers show up.

So when we look at SaaS, maybe we should ask something slightly sharper:

If this product became free tomorrow, would users still use it? Or would they quietly stop showing up because the incentive engine disappeared?

And if that incentive engine is the real service being sold, not just content and not just AI, then I can’t help but wonder:

Is the recent sell-off in SaaS names like Duolingo actually justified, or are we pricing in an AI narrative while ignoring the moat of making humans do hard things consistently?