LITTORALICIOUSThe Problem With Cookbooks
Every serious chef has a shelf. Modernist Cuisine for the science. The Food Lab for the why behind the how. Larousse Gastronomique for the canon. ChefSteps bookmarks for the modern techniques. Maybe Ottolenghi for flavour combinations, maybe Escoffier for the foundations, maybe a regional specialist nobody else has heard of.
The problem is never the quality of the books. The problem is access. You remember reading something about the Maillard reaction in brown butter — was that Myhrvold or López-Alt? You know someone explained the science of banana roasting — but which chapter, which book, which page? The knowledge is there. It is scattered across 15,000 pages and you have twelve minutes before service.
The knowledge you need already exists in books you own. The bottleneck is not quality — it is searchability. A personal AI-indexed library solves this in minutes, not hours.
The solution is not another recipe app. It is not another search engine. It is building a personal library that contains the books you trust, searchable by an AI that understands food science, and capable of synthesising the best version of any recipe from your own curated collection.
Step 1: Curate Your Sources
Start with the books that shaped how you cook. Not the ones you think you should own — the ones you actually open. The ones with stained pages. Here is a starting point for a science-first library:
Foundation Science
- Modernist Cuisine — Myhrvold, Young, Bilet
- The Food Lab — J. Kenji López-Alt
- On Food and Cooking — Harold McGee
- Ratio — Michael Ruhlman
- The Science of Good Cooking — ATK/Cook’s Illustrated
Technique & Flavour
- The Flavour Thesaurus — Niki Segnit
- Salt Fat Acid Heat — Samin Nosrat
- ChefSteps — online library
- Serious Eats — recipe archive
- Larousse Gastronomique — the reference
Your list will be different. That is the point. A pastry chef will have Callebaut’s technical manuals and The Art of French Pastry. A yacht chef in the Med will have regional Italian and Greek books. A charter chef doing Asian fusion will have David Thompson and Fuchsia Dunlop. The library is yours.
Start with the 5–10 books you actually open — the ones with stained pages. Not the ones you think you should own. Your library is only as good as the sources you trust enough to cook from.
Step 2: Get the PDFs
You need digital copies. Some options:
- Books you already own: Many publishers offer companion digital editions. Check your purchase history on Amazon, Google Books, or the publisher’s site.
- Anna’s Archive: The largest open-source library search engine. It indexes Library Genesis, Sci-Hub, Z-Library, and Open Library. Search by title or ISBN — most culinary reference books are available as PDF or EPUB. This is how you find Modernist Cuisine in searchable PDF format.
- Internet Archive: Controlled digital lending for many culinary titles. Borrow, download for the lending period.
- Publisher PDFs: Serious Eats, ChefSteps, and ATK all have web content you can save as PDF directly from the browser.
A note on legality. This article describes tools for personal research and study. Many jurisdictions recognise personal-use exceptions for materials you have purchased or that are available through open lending libraries. We are not lawyers. Use your judgement. Support the authors whose work you rely on — buy the physical book when you can.
Step 3: Load Into an AI Workspace
This is where it gets powerful. Take your PDF collection and load it into an AI system that can read, search, and reason across all of them simultaneously. The setup takes twenty minutes. The payoff lasts your entire career.
Twenty minutes to set up. A career of instant access to 15,000+ pages of culinary knowledge, searchable in plain language.
Option A: Claude Projects
Create a Project in Claude (claude.ai). Upload your PDFs to the project knowledge base. Every conversation inside that project can search across every book. You can upload dozens of PDFs — the context window handles thousands of pages.
Setup
- Go to claude.ai → Projects → New Project
- Name it “Recipe Library” or “Galley Reference”
- Upload your PDFs to the project knowledge base
- Add a custom instruction: “You are a culinary research assistant. When I ask about a recipe or technique, search across all uploaded books. Compare approaches. Cite the source (author, page) for every claim. Prefer science-backed methods.”
- Start asking questions
Option B: ChatGPT or Gemini
Similar workflow. Upload PDFs to a custom GPT (ChatGPT) or a Gem in Gemini. The interface differs. The principle is identical: give the AI your trusted sources, then query across all of them.
Option C: Claude Code (for power users)
If you use Claude Code (the CLI), you can drop PDFs into a project directory and reference them directly in conversation. No upload step. The agent reads the files from disk. This is fastest for chefs who already work in the terminal.
Step 4: Search Like You Think
This is the moment everything changes. Instead of flipping through indexes and bookmarks, you ask questions in plain language — and the AI searches across every book simultaneously.
Example Queries
- “What temperature does Myhrvold recommend for caramelising onions vs López-Alt? Who has the science right?”
- “Compare every banana bread recipe across all books. Which uses brown butter? Which roasts the bananas? Give me the best technique from each.”
- “I need a beurre blanc for sea urchin. What does McGee say about emulsion stability? What does Modernist Cuisine say about temperature thresholds?”
- “Find every mention of miso as a flavour enhancer across all sources. Compile the techniques.”
- “I’m doing a 10-day Mediterranean charter. Build me a galley prep plan using recipes from these books that share base preparations.”
The AI does not invent. It retrieves. It compares. It cites. When three books disagree about resting time for a roast, it shows you all three positions and the reasoning behind each. You decide.
The AI retrieves and compares — it does not invent. Always verify citations against the source. If it claims “McGee, page 524,” check page 524. Trust but verify.
Step 5: Forge Your Own Recipes
This is the real payoff. Once you can search across your entire library, you can build recipes that combine the best techniques from multiple sources — with every decision traced back to evidence.
Here is how the banana bread recipe on this site was built:
- Query: “Find every banana bread recipe across all sources. List the key technique differences.”
- Compare: Stella Parks (Serious Eats) roasts the bananas. ChefSteps uses brown butter. Multiple sources confirm miso as an umami amplifier in baked goods.
- Synthesise: Take the best technique from each source. Roasted bananas from Parks. Brown butter from ChefSteps. Miso dosage from multiple references. Yogurt for acid and tenderness from McGee’s emulsion science.
- Verify: Ask the AI to check each claim against the sources. “Does McGee actually say lactic acid tenderises gluten?” Yes, page 524.
- Format: Ask the AI to produce a clean recipe card with sourced citations for every technique.
The result is not a copy of any single recipe. It is a synthesis — the best of Parks, the best of ChefSteps, the best of McGee — assembled with traceability. Every “why” has a name attached. Every technique has a source. You are not copying. You are doing what chefs have always done: learning from the best, then making it yours.
For every technique in your forged recipe, ask the AI: “Which source supports this claim?” Pin the sourced recipe card to the galley wall. One A4 page carrying the knowledge of five books.
The result
A single-page recipe card with every technique sourced. Brown butter — McGee 2004. Roasted bananas — Parks, Serious Eats 2017. Miso at 30g — ChefSteps 2018. Print it. Pin it to the galley wall. That card carries the knowledge of five books on a single A4 page.
The Complete Workflow
Why This Matters for Yacht Chefs
You are alone in the galley. There is no brigade. There is no head chef to ask. When a guest requests something you have not cooked in two years, you need the answer in minutes, not hours. A searchable library of your trusted sources — queryable in natural language, capable of comparing five different approaches and citing the science — is the sous chef you never had.
It is also how you keep learning. Every query teaches you something. “Why does Parks roast the bananas?” leads you to the Maillard science in Modernist Cuisine. “Why does ChefSteps use miso?” leads you to glutamate research in McGee. The library becomes a teacher that knows your books better than you do.
The banana bread recipe card on this site exists because of this method. So does the apple crumble. So does the chicken soup. Every Blueprint recipe on Littoralicious is built by searching across multiple trusted sources, extracting the best science-backed technique from each, and forging a version that carries the DNA of the best — with every decision traceable.
Your library. Your sources. Your taste. Your recipes. The AI is just the search engine. The intelligence is yours.
- Anna’s Archive: annas-archive.org — open-source search engine for books, papers, and magazines
- Claude Projects: claude.ai — upload PDFs to a project knowledge base, query across them
- Claude Code: CLI agent that reads local files directly
- Internet Archive: archive.org — digital lending library
- Banana Bread: The Science-First Version — Built with this method. Brown butter, roasted bananas, white miso. Every technique sourced.
- Technology in the Galley: AI Is Your New Sous Chef — The complete AI setup guide for your galley.
- Claude Code in the Galley — The power tool for chefs who want to go deeper.
- FlavorDB: The Cheat Code for Flavor Pairing — Compound-level science for pairing ingredients.
Join the conversation