How to Import Recipes from Websites, Photos & Cookbooks.

Most people have more recipes than they can cook, scattered across more places than they can count. Browser bookmarks they never open, Instagram saves they can't find, screenshots buried in a camera roll, cookbook pages they photographed once and never looked at again. The problem isn't a lack of recipes — it's that they're never in the same place when you need them.

A digital recipe library solves this not by finding you more recipes, but by making the ones you already have actually usable. One searchable place, accessible from your phone, available whether you're planning on a Sunday afternoon or standing in the supermarket at 5pm.

This guide covers the four practical ways to import recipes into a digital library, what to look for in a recipe saving app, and how to build a collection that stays useful rather than becoming another pile of clutter in a different location.

The four ways to import a recipe

Recipes arrive in your life through different channels — a cooking website, a cookbook, a text your mum sent, a handwritten card from a friend. A good recipe library handles all of them. These are the four methods that cover every source:

Which method should I use?

Start with URL import for anything you find online — it's instant and accurate. Use photo import for cookbooks and physical cards. Use text paste for anything that arrives via message, email, or social media. Use manual entry for recipes that don't exist in any other form, or to record your own variations.

AI photo scanning: what actually happens

Of the four import methods, photo scanning is the one people are most sceptical about — and the one that most reliably surprises them. Here is what happens when you photograph a cookbook page, and how to get the best result.

What the AI reads

When you photograph a recipe, DinnerFlow sends the image to Gemini AI, which reads the text in the image and converts it into structured recipe data. It identifies the title, ingredient list with quantities and units, and the numbered instructions — returning them as editable fields you review before saving. This works on:

• Cookbook pages — any printed cookbook, whether the text is large-print or small-column
• Printed recipe cards — cards from recipe boxes, magazines, or newspaper clippings
• Handwritten recipes — family recipes on index cards or loose paper, including imperfect handwriting
• Screenshots and images — if a recipe appears as an image rather than text (a social media screenshot, a frame from a cooking video), photo import can extract it where URL import would fail

How to get a clean scan

Accuracy improves significantly with a few habits that take no extra effort once they're second nature:

• Fill the frame. The closer the page fills the camera view, the larger the text appears and the more accurately the AI reads it. Don't photograph a spread from arm's length with the recipe occupying half the frame.
• Even lighting, no glare. Sharp overhead light creates glare on glossy cookbook pages. Soft, even light — or just moving away from a direct window — gives cleaner results.
• Keep the page flat. A curved spine distorts the text at the edges. Press the book open firmly, or photograph one page at a time if the spine won't lie flat.
• One recipe per photo. If a spread contains two recipes, photograph them separately. The AI will attempt to extract whichever recipe it encounters first on the page.

Reviewing and editing after the scan

The extracted recipe always appears as editable fields before it's saved — you're not committing to whatever the AI produced. For most scans of well-printed cookbooks, the extracted recipe needs no correction. For handwritten notes with idiosyncratic shorthand or unusual abbreviations, a quick review takes under a minute. Correct any misread quantities, fill in a unit the AI left ambiguous, or fix a step that didn't parse cleanly — then save. The recipe is stored in your library exactly like anything imported by URL.

Building a recipe library that actually stays useful

Importing recipes is easy. Building a library you'll still want to use in six months takes a bit more intention. Three habits that make the difference:

Import as you go, not in batches

The temptation is to do a big one-off import of everything — every bookmark, every screenshot, every cookbook you own. That works, but a library of 200 recipes you imported in one sitting is harder to curate than one you built recipe by recipe. Import things as you come across them, and you'll naturally end up with a collection of recipes you actually want to cook.

Keep it curated, not comprehensive

A recipe library of 40 recipes you love is more useful than one with 400 you might cook someday. The goal isn't to digitize everything — it's to have a reliable rotation you can draw a meal plan from without having to scroll past dozens of things you've never made. When a recipe doesn't work for your household, delete it.

Import before you need the recipe, not during

The worst time to import a recipe is when you're already cooking. Import recipes whenever you come across them — while browsing, reading, or watching — so your library is ready when you sit down to plan the week. A few seconds of friction at import time saves you the scramble of remembering where you saw that pasta recipe.

Common questions, answered.

How DinnerFlow handles all four methods.

DinnerFlow is a free recipe planner app for Android built around the idea that importing a recipe should take less time than it takes to decide you want to save it. All four import methods are available in a single screen:

Once your library is built, DinnerFlow connects it directly to a weekly meal planning calendar and a shared grocery list — so importing a recipe is the first step in a chain that ends with a checked-off shop, not another bookmark you'll lose track of.