Paying someone to speak to you is going to provide comprehensible input and output.

Isn't individually targeted comprehensible input a large part of a 1-on-1 class? I would expect a good teacher to speak a lot during those 2 hours, and to properly adjust his speech so that's always pushing the boundaries of what the student can understand.

That approach feels intuitively wrong to me[1]; people who watch sports don't automatically become good athletes. People who look at pictures don't automatically become good at drawing. People who read don't automatically become good authors. People who watch cooking shows on TV don't automatically become good at cooking. Students who watch programming videos notoriously don't automatically become able to code anything the compiler accepts. Reviewing study notes by re-reading them is one of the less effective study strategies, compared to flashcards which prompt you to recall and generate answers from your memories.

Surely yes you need to adjust to the sound of a foreign language, but with no feedback loop of trying to speak and having another person feedback, how do you adjust?

Listening to hours of completely foreign language won't make you understand what the words are or what they mean, so "comprehensible" input includes weasel words that require you to already know the language before you can learn it. It's all over a bit weird.

Don't people say some of the most effective ways to learn are the immersion courses where you go to a retreat and speak only that language for weeks at a time, studying and learning 8 hrs+/day and then there are people who spend years reading or listening but aren't confident to speak anything. But are there people who spend years speaking with fluent speakers who still report they don't know the language?

[1] inb4 "hurr think you know better than professional linguists"

Check out the studies and lectures by Stephen Krashen if you want to hear from an academic that's been studying this for 40 years.