The AI Co-Reader for Philosophy

What is an AI co-reader?

An AI co-reader is an AI reading tool that ingests the entire book or document first, builds a structural model of it, and only then walks one passage at a time with that model in memory. The name is deliberate: a co-reader is not a summariser, not a retrieval engine, not a chat-with-PDF. It is a reading companion — the kind of colleague who has already read the book and is now sitting next to you with a cup of coffee while you work through the passage that stopped you.

That definition does real work. When you highlight a sentence in Heidegger's Being and Time and ask what it means, the answer should already know what Heidegger has established in the first six sections, which opponent he is arguing against in §7, and how this sentence cashes in a move he will reuse thirty pages later. That is not retrieval. That is structural reading, and it is what a co-reader is for.

The four categories of AI reading tools

1. General AI chat (ChatGPT, Claude, Gemini in a blank window). You paste a paragraph; it answers. The model has no idea whether your paragraph is the conclusion of a three-hundred-page argument or a throwaway illustration. For a recipe this is fine; for Kant's Critique of Pure Reason, where "intuition" was technically redefined forty pages earlier, this is a trap.
2. Chat-with-PDF tools (ChatPDF, Humata, Perplexity's PDF mode). The model retrieves passages by keyword match from the document you uploaded. A genuine improvement over general chat, but most implementations still treat the document as a retrieval index, not as an argument. They find the sentence containing your keyword and return it. They do not model inferential structure.
3. In-reader AI (highlight-and-ask features built into e-readers and PDF viewers). Smooth experience, but the underlying model is usually a small context window around the highlighted passage. The book is decoration; the model behaves like a general chat.
4. AI co-reader (Scholia). Ingests the whole book first, builds a structural model, then walks one passage at a time with that model loaded. The distinction is not cosmetic — it is the difference between retrieval and structural reading.

What Scholia actually does

Three concrete reading moves, each of them an engineering choice rather than a marketing claim.

1. Full-book context, always loaded

When you upload an edition — PDF or EPUB, the exact file you are reading — Scholia reads the entire thing before it answers a single question. That means Stephanus pagination in Plato, A/B pagination in Kant, §-numbers in Heidegger, and original-language terms (Dasein, epoché, différance, ousia) stay attached to the passage you are stuck on. Nothing is flattened into an English paraphrase that hides the edition it came from.

2. LAND before LIFT

Every Scholia response starts by echoing the exact phrase you tripped on — the surface of the sentence, in its original form. Only after that surface is in view does Scholia pivot to the mechanism: what the author is doing, what earlier move they are cashing in, which opponent they are answering. This is the opposite of a summary, which races to the paraphrase and loses the reader the specific sentence they needed help with. Cognitive science calls the summariser's failure mode the "fluency illusion" — the feeling of comprehension without the substance. LAND-before-LIFT refuses that shortcut.

3. Three-pillar frame: Skeleton, Environment, Soul

Skeleton asks what the load-bearing inference is — the argument's spine. Environment asks what the author's first audience already knew that you don't — the historical pressure the passage answers. Soul asks what problem the author was trying to solve when they wrote this sentence — the question that kept them at the desk. Three passes, held simultaneously, for every passage you walk.

Who Scholia is for

• Philosophy graduate students working through primary texts in multiple traditions
• Dissertation writers who need to cite primary sources with §-markers, A/B pagination, and original-language terms intact
• L2 (non-native English) readers navigating canonical hard books where the English translation loses the original's charge
• Autodidacts tackling Being and Time, Phenomenology of Perception, Dialectic of Enlightenment, Capital, or the Republic on their own
• Reading groups who want a companion that walks the passage with them rather than handing out spoilers

Where the blog comes in

The Scholia blog is the long-form demonstration of this reading posture. Every post walks a specific passage — Plato's Cave, Husserl's epoché, Adorno's culture industry, Merleau-Ponty's phantom limb — with the full-book context the method requires. If you want to see what "AI co-reader" means on an actual text before you upload your own, start there.

Featured reading guides

• Dasein Is Not 'Human Being' — Heidegger's German, the English loss
• How to Read Phenomenology of Perception, chapter by chapter
• Husserl's epoché: the bracket, not the dismissal
• What Foucault actually means by 'discipline'
• Plato's Cave is not the point of Book 7

Frequently asked questions

What is an AI co-reader?

An AI co-reader is an AI reading tool that ingests the entire book or document before answering any question about a specific passage. It then walks one passage at a time with the full-book structural model already in memory — so when you ask what a sentence means, the answer already knows what the author has established, what they are about to do, and which opponent they are arguing against at that exact moment. This is the opposite of a summariser, a chat-with-PDF tool that retrieves keyword matches, or a general AI chat that only sees the paragraph you pasted in.

How is Scholia different from ChatPDF?

ChatPDF treats the document as a retrieval index: it finds sentences containing your keyword and returns them. Scholia treats the document as an argument: it reads the whole edition first, builds a structural model of it, and then walks one passage at a time with that model loaded. For a philosophy text where the same word can carry three technical senses across three chapters, the retrieval approach misses the move. Scholia is built for the books where the argument lives in the architecture, not any single paragraph.

What is the LAND-before-LIFT method?

LAND-before-LIFT is the core reading move Scholia uses. LAND means echoing the exact phrase the reader tripped on, in its original form, so the reader recognises it as the same sentence they were reading. LIFT means pivoting from that surface to the mechanism: what the author is doing with that phrase, what earlier move they are cashing in, what argument they are building. Summarise-first tools skip LAND entirely and go straight to a paraphrase, which feels fluent but loses the reader the specific sentence they needed help with.

Who is Scholia for?

Philosophy graduate students, humanities researchers, dissertation writers, L2 (non-native English) readers working through canonical texts, and autodidacts tackling hard books on their own. If you are reading Heidegger, Husserl, Merleau-Ponty, Foucault, Adorno, Plato, Kant, Derrida, Butler, Deleuze — or any text where the argument is dense enough that a summary misleads you — Scholia is built for you.

Does Scholia summarise the book?

No — deliberately not. Summaries produce the 'fluency illusion' — the feeling of comprehension without the substance — and that illusion is exactly the failure mode Scholia is designed to avoid. Scholia walks one passage at a time and asks you to stay with the text. It is a companion for the reading, not a replacement for it.