In April 2025, People Inc did something quietly significant: it launched MyRecipes, a single platform aggregating all 100,000+ recipes across Allrecipes, Food & Wine, Martha Stewart, EatingWell, and Real Simple. By late 2025, it had 2.1 million registered users. The company projected tripling that by the end of 2026.
In October 2025, People Inc acquired Feedfeed, a creator-driven food community. Content plus community. Owned audience. Direct relationship.
This is the right product instinct. If Google AI Overviews answers “how do I make chicken piccata” directly in the SERP, the recipe page is dead. But the logged-in user who saves that recipe to their collection, exports the ingredient list to Kroger, and comes back next Thursday for something new — that user can’t be intermediated. MyRecipes is the post-search recipe product.
The instinct is right. The ingredient list is incomplete.
What Raptive Built
On June 9, 2026, Raptive launched Raptive Intelligence and simultaneously acquired AlchemyAI, a stealth agentic food intelligence platform. John Roa, AlchemyAI’s founder, became Raptive’s Chief AI Officer.
The acquisition brought a food knowledge graph that maps relationships most recipe databases don’t even model: recipes to ingredients to products to preferences to techniques to substitutions to nutrition to shopping intent. Not a flat database of 100,000 recipe pages. A relational graph of how food decisions actually happen.
Here’s Roa’s pitch: “All of the underlying data that would be invisible from walking across the Internet is going to be one of the most valuable assets in the data ecosystem.”
He’s right. Raptive has 224 million monthly users in food and lifestyle — #1 in food on Comscore for seven consecutive years, outranking Dotdash Meredith. But the competitive advantage isn’t the reach. It’s the behavioral data. Raptive knows what those 224 million people save, modify, cook, substitute, and shop. Not what editors published — what users did with it.
The business model is data licensing to CPGs, retailers, and AI platforms — not advertising. Three revenue streams: licensed data access, real-time intelligence products, and agentic commerce infrastructure. Revenue shared back to creators, which keeps the flywheel turning.
Raptive Intelligence is not a publisher. It’s a data company that happens to sit between 6,500 food creators and every grocery buyer in America.
What Amazon Signed
In May 2025, Amazon signed a multi-year AI licensing deal with The New York Times worth $20–25 million annually. The deal covers NYT articles, NYT Cooking recipes, and The Athletic — all feeding into Alexa and Amazon’s shopping infrastructure.
Six weeks later, Condé Nast and Hearst signed similar deals. Bon Appétit. Epicurious. Delish. Every major food publisher’s recipe corpus, licensed to power what Amazon now calls Alexa for Shopping — a unified agentic commerce assistant merging Rufus AI with Alexa+ across mobile, desktop, and Echo Show.
Amazon now offers same-day delivery of perishable groceries in 2,300 U.S. cities. Perishable grocery sales grew 40x in 2025. The pipeline from “What’s for dinner?” to a filled cart on your doorstep in four hours is not theoretical. It’s live.
Allrecipes is absent from every one of these deals.
People Inc had an AmazonFresh/Allrecipes integration in 2018 — shoppable ingredient buttons embedded directly in recipes. It predates the current AI licensing wave. Amazon has since closed its Fresh physical stores, pivoted to delivery, and signed content licensing deals with everyone except the company that owns the world’s largest recipe site by traffic.
Whether that’s a negotiation in progress, a strategic holdout, or a gap that hasn’t been noticed — from the outside, the result is the same. The recipe-to-cart pipeline is being built without Allrecipes in it.
The Capability Map
■ Operational ■ In progress ■ Gap
People Inc leads on content. It has the recipes. It has the brand trust. It’s building the registered user base. But move down the stack — knowledge graph, fulfillment, protocols — and the gaps open up. The top two rows are green. The bottom three are red.
Three Things the Engineering Team Should Build
1. A food knowledge graph, not a content taxonomy.
D/Cipher classifies content. It knows that a page is about “chicken piccata” and maps it to a contextual taxonomy for ad targeting. That’s valuable for programmatic. It’s useless for commerce.
Commerce needs a different kind of intelligence: this recipe calls for 8 oz of thin-cut chicken breast, which maps to these SKUs at these retailers, at these current prices, with these substitutions for dietary constraints. The user who saves this recipe also saved three pasta recipes last week, suggesting a weekly meal-prep pattern that a grocery partner could fulfill with a single delivery.
This is what AlchemyAI built. It’s the layer between content and commerce that D/Cipher doesn’t occupy. Graphene-Selene has the content model. The ingredient-to-product mapping, the substitution logic, the intent inference — that’s what’s missing.
2. Grocery retailer API integrations, not export buttons.
“Export shopping list to retailer” is a PDF. A real commerce integration is a bidirectional API: real-time inventory feeds from Kroger, Instacart, Walmart, Amazon Fresh. Current prices. Availability by zip code. Substitution suggestions when the store is out of tarragon. Cart pre-population with one tap.
The 2018 AmazonFresh integration was a version of this. It was early and limited. The grocery delivery infrastructure has since grown 40x. MyRecipes needs retailer integrations that match the current scale — not a single partner, but a multi-retailer fulfillment layer that lets the user choose.
3. A protocol surface for agentic discovery.
When an AI agent asks “What should I make for dinner tonight that uses the chicken in my fridge and takes under 30 minutes?” — what answers? Right now, the agent queries whatever content its LLM has been trained on, or whatever publisher has an MCP server exposing structured recipe data.
MyRecipes has no documented API surface. No MCP server. No way for an external agent to discover, query, filter, or transact against 100,000 recipes. The same closed-platform pattern that keeps D/Cipher out of the agentic ad stack keeps MyRecipes out of the agentic commerce stack.
Raptive is positioning its knowledge graph as a data layer above the protocol wars — protocol-agnostic, selling intelligence to whatever agent asks. Amazon is building Alexa for Shopping as the agent itself. MyRecipes needs to decide: be the data layer, be discoverable by agents, or be invisible to the fastest-growing content distribution channel.
The Pattern
This is the bungalow again. The same organizational instinct — own the content, curate the experience, control the relationship — produces beautiful products and closed systems. MyRecipes is a beautiful product. It is also a closed system in a market where the transaction layer is being built by companies that treat content as an input, not a destination.
The ingredient list for making MyRecipes into a commerce platform: a knowledge graph that maps recipes to products, retailer integrations that connect intent to fulfillment, and a protocol surface that makes 100,000 recipes discoverable by machines. People Inc has the first ingredient — the recipes. The other three are on a shelf in someone else’s kitchen.