Reading summaries alone is insufficient for evaluating a book's worth.

A personal assistant that can show, not tell, turns vague book recommendations into something checkable before the first page is turned.

Asking an LLM for a reading list used to mean accepting a summary written by someone who had never held the book. Those summaries strip out everything that makes a voice distinct—the cadence, the tone, the way ideas unfold in real sentences. When the model finally gets a recommendation wrong, the cost is measured in hours of sunk reading time.

The Coffeetable connector for Claude replaces that guessing game. Once installed, the assistant does not describe a book from memory; it pulls actual pages from the text and slides them straight into the chat window. That single change moves the work from abstract description to concrete sampling.

The shift happens in three stages. First, a high-level prompt shaped by Claude’s long-term memory—something like “Based on everything you know about my interests, my current projects, and the way I think, what are some books I must read next?”—produces a shortlist that already carries personal context. Second, each title can be verified on the spot by requesting the real content instead of trusting a recap. Third, specific passages can be pulled for closer inspection, such as “Show me the first three pages of Chapter 2 so I can get a feel for the author's style.”

Prompt engineering for discovery now optimizes for sampling rather than description. Instead of refining how the desired book is described, the focus becomes refining which slice of the source material lands in the context window. That lets the reader validate the prose before committing, turning Claude from a librarian into a preview reader. If the sampled pages fail to engage, the book is shelved without wasting time.

For anyone building an LLM agent ecosystem, the lesson is clear: a model performs only as well as its grounding in real-world data. A few habits make the difference. Generic queries like “best sci-fi books” collapse into the same top-10 list every time, so ask for something specific instead—“Show me a passage from this book that demonstrates its core philosophy.” Pair that with note-taking so the sampled pages become immediate, high-quality study notes or summaries within the same thread.

Moving from “tell me about this” to “show me this” is not an upgrade; it is a reset.