The Future of Recipe Discovery: From Handwritten Cards to AI Extraction

My grandmother kept her recipes in a metal tin. Index cards, some typed, some handwritten in cursive that got harder to read as the ink faded. Every card was a complete recipe: title at the top, ingredients down the left side, instructions across the body. If you wanted her pot roast, you pulled the card, propped it against the backsplash, and cooked. The system was limited — you only had recipes you’d been given or copied by hand — but within those limits, it worked perfectly. The recipe was always findable, always readable, always structured.

That tin box was the last time recipe saving was a solved problem. Every era since has traded discoverability for usability, finding more recipes but losing the ability to keep them.

The cookbook era: curated but constrained

Printed cookbooks were the first expansion beyond personal recipe collections. Suddenly you had access to hundreds of tested recipes organized by cuisine, ingredient, or technique. A good cookbook was a reference manual you could trust.

But cookbooks imposed someone else’s taste. You cooked what the author chose to publish. Discovery was limited to what bookstores stocked and what friends recommended. If you wanted your aunt’s specific version of tamales, you still needed the handwritten card. Cookbooks also assumed a certain baseline of knowledge — “fold until just combined” is clear to an experienced baker and meaningless to a beginner.

Still, the recipe itself remained structured. Ingredients, quantities, steps. The format hadn’t broken yet.

The internet era: infinite recipes, fragile bookmarks

Food blogs and recipe websites blew the doors off discovery. Suddenly you had millions of recipes from every cuisine on earth, searchable by ingredient, dietary restriction, cook time, and rating. The constraint of the cookbook era — someone else choosing what you had access to — evaporated overnight.

The saving problem resurfaced in a new form. You bookmarked URLs. You pinned on Pinterest. You emailed links to yourself. Each method created a reference to the recipe, not a copy of it. When a blog went offline, your bookmark died. When Pinterest changed its algorithm, your pins got buried. When your browser bookmark folder hit 200 entries with no organization, finding the right recipe became its own project.

Pinterest deserves specific mention because it came closest to solving the problem. A visual grid of saved recipes, loosely organized into boards, searchable by the text in the pin. For a few years in the early 2010s, Pinterest was genuinely the best recipe-saving tool available. But it was still a layer of indirection — you pinned a link to a blog post, and if the blog post changed or disappeared, your pin became a thumbnail pointing to nothing.

The recipe format itself held, though. Blog recipes were still structured: ingredient lists, numbered steps, print buttons. The information was all there. The problem was keeping track of it.

This is where the story gets interesting and messy.

Social media transformed recipe discovery in ways that no one predicted. Instagram, TikTok, YouTube, and Facebook became the primary places where people — especially younger cooks — found new recipes. The shift brought real, meaningful improvements to food culture. Home cooks gained access to recipes from cuisines they’d never encountered. A teenager in Ohio could learn to make dumplings from a grandmother in Sichuan. A college student in Lagos could pick up a sheet-pan dinner technique from a chef in London. The diversity and accessibility of recipe content expanded enormously, and that’s genuinely good.

But social media was built for engagement, not for reference. The recipe stopped being a structured document and became a performance. On TikTok, the recipe is spoken aloud over 60 seconds of rapid-fire footage. On Instagram, it’s scattered across carousel slides and buried in a caption below a personal essay. On YouTube, it’s embedded in a 20-minute video where the actual cooking starts at the 8-minute mark.

The format that made recipes discoverable — short, visual, personality-driven content — is precisely what makes them impossible to save. You can’t search a video. You can’t scale the servings on a screenshot. You can’t check off ingredients in an Instagram caption.

This is the paradox of the social media recipe era: we have never had better access to more recipes from more places, and we have never been worse at keeping them.

The screenshot graveyard

Faced with this gap, people improvised. The dominant strategy became the screenshot — the most instinctive and least effective recipe-saving method ever invented.

Screenshotting a recipe takes a fraction of a second. It feels like you’ve captured something. But what you’ve actually done is convert a recipe into an unsearchable image buried in a camera roll alongside parking garage tickets and photos of your dog. There’s no title, no ingredient list, no tags, no search. Three months later, when you want to make that Thai basil chicken, you’re scrolling through hundreds of thumbnails hoping to visually recognize one.

For video recipes — which now account for the majority of recipe content on social media — screenshots are even worse. You can screenshot a single frame, but the recipe was spoken aloud across 45 seconds of footage. That screenshot captures maybe one ingredient list if you’re lucky. The instructions, the timing, the technique — all of it is still locked inside the video.

Meanwhile, the videos themselves keep disappearing. Creators delete posts, accounts go private, platforms take down content. Your bookmark to that incredible birria taco recipe now points to a dead link. The recipe is gone, and you never had a real copy.

The gap between discovery and preservation

This is the core tension in recipe culture right now. Discovery has never been richer. Preservation has never been more fragile.

Every previous era of recipe management had a natural pairing between how you found recipes and how you kept them. Handwritten cards were both the discovery mechanism and the storage format. Cookbooks were browsed and shelved. Blog recipes were found by search and saved by bookmark — imperfect, but at least the format remained structured.

Social media broke that pairing. The discovery format (short-form video, ephemeral stories, algorithmically surfaced content) has no relationship to any reasonable storage format. A 30-second TikTok is a terrible container for a recipe you want to cook six months from now. But it’s an incredible container for making you want to cook something new tonight.

The result is a growing pile of lost recipes. Every home cook has a version of this experience: you saw something amazing, you tried to save it, and when you went to cook it, the recipe was gone or unusable. The more recipes you discover, the more you lose. The gap between discovery and preservation widens with every scroll session.

AI extraction: the recipe becomes data again

This is where the future of recipe discovery points. AI extraction doesn’t ask you to change how you discover recipes — keep scrolling, keep watching, keep finding things that make you want to cook. It changes what happens after you find something.

When you share a URL to an AI extraction tool, the recipe is read, understood, and converted into structured data. Ingredients with quantities, parsed and normalized. Steps in order. Cook times, prep times, servings. The recipe becomes a recipe again — not a video, not a screenshot, not a caption, not a bookmark. Data. Searchable, editable, scalable, permanent data.

What makes this different from every previous saving method is that AI extraction is format-agnostic. It doesn’t matter whether the recipe arrived as a TikTok, an Instagram carousel, a YouTube tutorial, a food blog, a photo of a cookbook page, or a screenshot of a text message from your mom. The AI watches videos, listens to narration, reads text, and analyzes images to produce the same structured output regardless of source format.

This is the leap. For the first time, the preservation layer is decoupled from the discovery layer. You can discover recipes in whatever chaotic, unstructured, delightful format social media serves them up — and save them in a format that actually works for cooking.

What this looks like in practice

Pluck is built around this idea. The workflow is simple: you find a recipe anywhere on the internet, share the link or photo to Pluck, and the AI extracts a structured recipe in seconds. You review it, make any corrections, and save it to your recipe box.

Behind that simplicity is a multi-modal extraction pipeline that handles the real complexity. For video recipes, the AI samples key frames, transcribes audio, reads on-screen text, parses captions and descriptions, and cross-references all of those signals to produce the most complete and accurate extraction possible. Each extraction gets a confidence score so you know how much to trust it before you start cooking.

The recipe box itself is what the index card tin always should have been: searchable by ingredient, title, cuisine, or tag. Accessible offline. Editable when you want to tweak a recipe after cooking it. Structured so you can read it while your hands are covered in flour.

Looking forward

The future of recipe discovery isn’t about choosing between the richness of social media and the reliability of structured recipes. It’s about bridging the two — letting people discover food in whatever format sparks their appetite and saving it in whatever format serves them in the kitchen.

AI extraction is still improving. Video understanding gets more accurate every quarter. The models get better at handling fast speech, background noise, vague measurements, and the thousand small challenges of parsing real human cooking content. The gap between what you discover and what you keep is closing.

The grandmother’s tin box worked because the format was simple and permanent. The future works because the technology is finally smart enough to convert any format into something equally simple and permanent. The recipe card never went away — it just needed a translator.

Pluck extracts structured recipes from any URL, video, or photo using AI — so the recipes you discover actually become recipes you cook. Available now on Android — get it on Google Play. iOS coming soon; join the waitlist to be notified.