So I put together one that's only about books: drafting, continuity, editing, covers, formatting, KDP research.
[Github] Curated List of Awesome AI Writers: https://github.com/finxter/awesome-ai-book-writing
There's also a section on workflows that actually work, including the fix for voice drifting over long books. PRs welcome if I missed something obvious.
If you have ever generated a long book and noticed the last third reads like a different, blander author wrote it, that is not you imagining things. It is long-context style decay. The model honors your voice instructions early, then quietly drifts back to its house style as the context fills up.
Two writers flagged it to me the same week, both with the same shape: book one great, later chapters mushy, same prompt. So it was not the prompt.
The fix that actually worked: stop putting your voice and style instructions only at the top of the job. Re-inject them at the start of every single chapter, along with a couple of paragraphs of your own prose as a sample. Costs almost nothing in tokens. The difference on an 80k-word draft was night and day, and one of the writers confirmed it held to the last page.
If you are drafting chapter by chapter with Claude or GPT, this is a five-minute change to your loop. If you are using one giant prompt, this is your sign to stop doing that.
If you use a platform such as ImagineYourBook, this fix might already be implemented (it is for IYB).
Anyone found other tricks for holding voice over long generations? Curious whether style samples beat instructions in your experience, because in mine they did by a lot.
I always love a good old Reddit tool thread.
So what is the best ai tool to write full books (end-to-end, prompt-to-book)?
- Romance
- Biographies
- Non-Fiction
- Children books
- ...
I have been happily using ImagineYourBook a lot lately but I guess it doesn't harm to keep looking around.
For my sources of passive income I always had to work very hard.
Say I publish a book on Amazon to generate passive income. For ease of math, say I work 40h total to create and publish the book and make $100/m on average for 36 months.
While this might be satisfactory to some people, it's just an above-average active income when looked at it like this. And it's higher than average because there's no certainty and high volatility. You have to put in the hours before you know how much you'll make. Most won't do this.
So in effect, striving for passive income (in my experience) is a good life choice because of delayed gratification and higher average hourly rate due to less competition. But it's not per-se passive, only delayed.
That's the post. Let's go!
I was always curious if we can get the number of sales from the number of reviews (roughly).
How many units do you need to sell to get to one review?
I'm aware that ppl fake reviews that's why I'm asking here hoping to get some honest answers.
For me, it's roughly 20 units sold per 1 review for one of my top book