Not a transcript read aloud. It is a two-person show: the book's author "joins" as a guest, and an AI host interviews them, with every question drawn from your role and what you need from the book. Same book, different episode for a PM and a designer.
I kept running into books people around me swore by, and I can listen to podcasts, so the obvious move was to turn books into audio.
Except every existing tool (including NotebookLM) generates the same episode for everyone, with generic examples that don't map to what I actually deal with at work. A PM running an AI agent for auto dealerships and a designer at a fintech startup should take completely different things away from the same book.
So I built one that knows who I am, and stages the book as an interview: the host asks the questions I would ask, and the author answers.
The EPUB is parsed into clean chapter text. TOC, copyright, and blank pages are filtered out automatically.
Each chapter is summarised with your profile injected, so the digest emphasises what matters to you.
Digests are woven into a two-voice interview script (host + author). Questions and examples are personalised to your role.
The script becomes multi-voice audio via neural TTS, batched asynchronously, then stitched into one MP3.