Everyday

Never Wonder What to Read Next: Let AI Keep Your Reading List and Find Your Next Series

The problem

Reading is one of the ways I unwind. The hard part is what comes after a great book or series: figuring out what to read next. I look for books with a good reputation, but "best books of the year" lists are written for everyone, not for me, and sorting through reviews takes time.

And after years of reading, I can't always remember what I've already read. More than once I've been 20 pages into a book and thought, "Wait a minute, I've read this." It's annoying.

A Kindle e-reader in an open black cover on a wood table, its screen in dark mode showing a page of text
My Kindle, where most of my reading happens. (I swapped the book page on the screen for the opening of this project.)

So I gave the job to AI. Claude keeps track of what I read and what I thought of it, and when I'm ready for something new, it suggests books and series that fit my taste and tells me why. Some of my favorite series came from those suggestions. I was reading Dungeon Crawler Carl before it got big, because the AI recommended it. I found the Bobiverse books, which I probably wouldn't have tried on my own. And it pointed me to Expeditionary Force, a series of more than twenty books with big world-building, humor and characters who grow along the way: everything I like.

What you need

How it works

  1. A reading folder in your private repository holds two things: a list of every book you've read, each marked liked or disliked, and a short taste profile that says what you enjoy and what you don't.
  2. You talk about books. "I finished the third one, loved it, the ending was a little rushed." Claude adds it to the list with your note. No typing into forms.
  3. When you want something new, you ask. Claude reads your list and your taste profile first, so it doesn't suggest books you've read or styles you've told it you don't like.
  4. You point it at your sources. Tell it where you get books, and it leads with those. Mine leads with Kindle Unlimited, so I can start a whole series the same day without buying anything.
  5. Every pick comes with a reason. For example: "same dry humor and clever problem-solving as the books you rated highest." It also says whether a book starts strong or takes a while, and how long the series is.
  6. It gets better over time. Each like and dislike you add sharpens the next round of picks.

My taste profile is a few lines. I like science fiction and fantasy with fast plots, smart heroes who solve problems creatively, humor mixed with real stakes, and long series that pay off. I don't like dense, slow writing, bleak endings with no payoff, or stories where nothing happens for a long time. That's enough for the AI to see patterns I wouldn't have put into words, like how much I enjoy a witty AI character or a hero who's simply very good at their job.

It started as a small program on my home computer. My first book agent ran on a Mac mini once a week and texted me its picks. Once my private repository was keeping my reading list and preferences up to date on its own, I moved the agent to a Claude scheduled task. Now there's nothing to maintain at home, and every week's picks start from my latest list.

Setting it up

  1. Set up the private repository first, if you haven't already. Teach AI About Your Life Without Handing It Over walks through it, including the privacy rules and the "push" keyword.
  2. Create the reading folder and your list. Start a session with your repository selected and paste this:
Create a folder for my reading. Add a books-read list with title, author, series, the date I finished it if I know it, and whether I liked it, with a short note in my words. Add a taste profile that sums up what I like and dislike in books. I'll tell you the books I remember; ask me a few questions about each so the list is useful for recommendations later. Show me everything before saving, and save when I say push.
  1. Talk through the books you remember. Favorites, books you gave up on and series you never finished are all useful. The dislikes matter as much as the likes.
  2. Ask for recommendations:
Read my books-read list and taste profile, then recommend books and series I'm likely to enjoy. Don't suggest anything already on my list. Favor books and series with a strong reputation among readers. Lead with books included in Kindle Unlimited, and say which picks are and aren't. Group them as close matches, stretch picks and hidden gems. For each one, give the title, author, a short pitch with no spoilers, why it fits my taste, whether it starts strong or takes time, and how long the series is and whether it's finished. Keep each pick under 150 words, and end with the three I should start first.

Change "Kindle Unlimited" to wherever you get your books, such as your library's ebook app, or leave that sentence out.

  1. Check a pick before you start it. Confirm it exists, see the series order and check that it's really in your subscription or library. Catalogs change, and the AI can be out of date.
  2. Put it on a schedule. Set up a weekly scheduled task with the recommendation prompt, and turn on notifications so the picks come to you. That's how I run mine.

Upkeep

What it doesn't fix

Where to start: the last book you loved and the last one you gave up on, with a sentence on why. That's enough for a first round of picks, and the list grows from there.

Products named here are ones I bought and use myself. No sponsorships, no affiliate links.
I'm not a licensed electrician or other professional. Follow the product instructions and local codes, and call a qualified pro if you're unsure. Full disclaimer.

← All projects