The AI has guessed how the book is divided. Correcting that guess here costs nothing; every lesson you leave in the list costs one AI call when you press Read N Lessons.
Read down the units and lessons.
A unit holds several lessons. A lesson is one study. "What generating costs" beside them counts what is to read, what is skipped and what is already read.
Skip anything that is not a lesson.
A foreword, a contents page, an answer key - Skip on a lesson, Skip All on a unit, Include to bring one back. Skipped sections are never read and never reach the library, so skipping is the cheapest correction there is. The bin deletes a unit or lesson from the list outright.
Check the Book panel.
Edit it to correct the title, author and description - and, for a new series, its name - then Done.
Rename anything with the pencil, and move it with the arrows.
An arrow past the top or bottom of a unit moves the lesson into the next unit along.
Add a unit if the book needs one it did not get.
Press Add Unit, rename it, then walk lessons into it with the arrows.
Fix a page range if it looks wrong.
Pages is where in the PDF the lesson lives, and it is what the AI is told to read from - a wrong range makes it read the wrong pages.
Press Read N Lessons, and it reads them one at a time.
Each row goes Queued, Reading, Done - or Failed, with an i that says why. Stop After This Lesson pauses the run; Read the Remaining N carries on, and See the Content Model moves on once it is done. With nothing left to read the button is Continue to the Content Model.
A unit holding a single lesson is usually wrong - it means the book's sections ARE the lessons. A twelve-week book with one study a week is twelve lessons, not twelve units. The screen says so when it sees it, and offers to merge them in one press.
Discard Import throws the whole import away at once, without asking.
Introduction and Conclusion are the exception: one lesson each is exactly right, and how these books are built by hand.