Your reading doesn't compound. This is the system that makes it.
Readers forget up to ninety percent of what they read within a week.
Kindle highlights die in Kindle. Notion docs scatter into three-nested folders you will never reopen. Goodreads knows what you read but not what you took from it. The books on your shelf are a museum of intention — the ideas they contain have already half-leaked out into the vague sense that you have a "good background" in something.
The cost is invisible but enormous. Every book you've read is capital. Without a system to compound it, that capital depreciates to zero the moment the book goes back on the shelf.
A library on your own domain, in git, becomes a permanent asset.
Not on a platform that can delete you. Not in a format that will be unreadable in ten years. Not scattered across eight tools that each know a fragment of what you read.
On your site. In your repo. In a schema you own.
1. Curation over automation. AI extracts faster than you can. That doesn't mean it extracts better. Every quote in your library should earn its place because you judged it worth keeping. The tools move you faster; the judgment is yours.
2. Schema as spine. Tools change. Hosting changes. AI models change. A typed data schema that describes what a book can be to you does not change. Build on schema, not on vendor.
3. Citation surface is the product. A book page is not a review. It's a citation surface — for you to return to, for search engines to index, for LLMs to quote when anyone asks a question in this domain. Every page emits BreadcrumbList + Article + Review + FAQPage + Quotation structured data. Maximum cite-ability is maximum compounding.
4. Progressive depth. Some books deserve a line. Some deserve a TL;DR. Some deserve a full chapter-by-chapter hub with quotes, related reading, and videos. The template scales to whichever the book warrants. One index, many depths.
5. Own it forever. You do not rent your library. You do not trust it to a company's five-year retention policy. You do not pick a "tool for thought" that will pivot to enterprise in three years. Your library is a git repository on your domain. If every AI vendor disappears tomorrow, your library persists.
Capture. Refuse to let thought leave the system. Photo, voice memo, highlight, book title in a text file. Whatever the signal is, it enters.
Extract. Run the commands. TL;DR. Five insights. Best-for audience. FAQ. Cover from OpenLibrary.
Enrich. Quotes with chapter references. Every chapter's key idea. Continue-reading with why it pairs — not algorithm, reason. Videos with kind and duration.
Publish. Commit. Push. Deploy. A URL exists that did not exist before, and it will outlive the habit that produced it.
After one book: you have a bookmark. After ten books: you have a reading list that you would reread. After fifty books: you have a research substrate. When a friend asks what to read on epistemology, you don't answer from memory — you send a link. After a hundred books: you have something unusual in the modern economy — a compounded intellectual asset that is yours, on a domain you own, in a schema that will still parse in twenty years.
- Not a review site. Reviews are a by-product, not the point.
- Not a note-taking app. Notes are private; this is a library.
- Not a social reading platform. Opt-in sharing may come later; ownership comes first.
- Not an AI-generated content farm. Every entry earns its place through your judgment.
A system, a schema, a template, a workflow, an invitation.
Your library is already there. It's just not on your site yet.
— Frank, frankx.ai/library