Apps are good at things lessons are bad at, and bad at the one thing lessons exist for. Most people would be better off using both than arguing about which is better.
What apps genuinely do well
They build vocabulary efficiently. They make a daily habit almost frictionless — five minutes on a phone, no scheduling, no one waiting for you. They're cheap or free. And for absolute beginners they remove the first barrier, which is not knowing where to start.
For recognising words, drilling verb forms and keeping a language ticking over, an app is a genuinely good tool. Anyone who tells you otherwise is selling lessons.
What they can't do
Apps test recognition — you see four options and pick one. Speaking requires recall, where nothing is in front of you and the word has to arrive on its own. Those are different skills, and strength in the first doesn't produce the second.
Nothing on an app is unpredictable. Real conversation is interruption, a question you didn't expect, someone speaking faster than you'd like. You can't rehearse that against software that waits patiently.
And nothing corrects your register. An app will accept a sentence that is technically right and socially wrong — too blunt, too formal, not how anyone says it. Only a person tells you that.
The streak problem
Streaks measure consistency, not progress, and the two feel identical from the inside. It's entirely possible to maintain a two-year streak and remain unable to hold a conversation, because the thing being measured was turning up, not speaking.
If you've been doing an app daily for months and still freeze when someone addresses you, that isn't a failure of discipline. The tool was never doing the thing you want.
When an app is enough
If you want holiday phrases, or you're maintaining a language you already speak, or you're in the first few weeks and just building a base, an app is fine and lessons would be premature.
When it isn't
The moment your goal involves speaking to a person. At that point you need someone who makes you talk, interrupts, and tells you what you just got wrong — and no amount of additional app time substitutes.
The sensible combination
App for vocabulary and daily contact. Lessons for production and correction. One hour a week of speaking alongside the app you're already using moves faster than either alone, and considerably faster than another year of streaks.
If your question is specifically about AI chatbots rather than gamified apps, that's a different subject — see the companion piece on why AI can't replace a real language teacher.