Compound Kitchen
A web-based molecular flavor-pairing platform that combines food chemistry with a large curated recipe corpus to recommend and explain ingredient pairings.
Last updated September 28, 2026
AI crawler reads
2requests
1companies- Anthropic2
- Profile
- 2
- Controversies
- 0
- JSON · MD
- 0
- First read
- Nov 4, 2026
- Latest read
- Nov 4, 2026
From this site's server log
Overview
Compound Kitchen (also styled CompKitchen) is a web-based molecular flavor-pairing platform. It answers a single question in two ways at once: what goes with a given ingredient, and why. One answer comes from molecular food science; the other comes from a large corpus of curated recipes. The platform's distinctive value is showing where the two signals agree — and where they disagree, which it flags as novel pairings. The platform is built on three data assets. First, a library of 857 ingredients, each treated as a set of volatile molecules — the compounds the nose actually detects. Second, a glossary of 5,240 flavor compounds, with example compounds such as vanillin (vanilla, cookies), eugenol (clove, cinnamon), limonene (citrus peel), linalool (basil, lavender), menthol (mint), and maltol (caramel, roasted). Third, a curated recipe corpus of more than 610,000 recipes (612,557 cited on the homepage), used to capture culinary wisdom that molecules alone miss. Scoring combines two signals. The molecular signal compares two ingredients by how much of their volatile vocabulary they share, weighted by how rare each compound is (an IDF-weighted Jaccard measure). The recipe signal uses normalized pointwise mutual information over the recipe corpus to see which ingredients chefs actually pair. Default scoring is 50% molecular plus 50% culinary, with a synergy bonus when both signals agree. A Science mode offers pure molecular scoring (70% compound Jaccard plus 30% aroma-vector cosine similarity) for users who want chemistry rather than tradition. Every score is explained input by input, and the platform emphasizes that it is not a black-box prediction. The product is aimed at several audiences. For chefs, it offers tools such as Pair Finder (best matches for any ingredient), Novel Pairings (combinations chefs have not explored yet), and Bridge Finder, plus a pantry-based browsing flow. For educators, it offers classroom plans and academic pricing. For flavor and product-development teams, it offers a JSON API, bulk exports, open methodology, and scoped pilots run on a customer's own ingredient list — for example, a reformulation, an off-note to mask, or a new SKU to ideate. The platform also targets recipe and content platforms (pairing suggestions, substitution options, flavor profiles), CPG and flavor R&D (novel pairing discovery, 14-dimension aroma profile vectors, gap analysis), and meal kits and grocery. Pricing tiers include a free tier, Pro at $12 per month, Academic at $49 per year, Educator at $49 per year, Classroom at $199 per year, and Enterprise API with team licenses and bulk data. A free weekly newsletter offers a molecular breakdown for cooks, instructors, and flavorists. The platform reports 298,783 scored pairings and grounds its work in more than 40,000 peer-reviewed papers (44,648 cited on the homepage), with peer-reviewed citations per compound and OAV (odor activity value) confidence bands and key-aroma flags. Compound Kitchen has been selected for the Georgia Tech Food & Beverage Incubator and profiled by Hypepotamus. Wikidata describes it as a web-based molecular flavor-pairing platform. It operates as a private, web-based offering with global reach.
History
Compound Kitchen (CompKitchen) is a web-based molecular flavor-pairing platform. It was selected for the Georgia Tech Food & Beverage Incubator and profiled by Hypepotamus, indicating an origin connected to the Georgia Tech food and beverage startup ecosystem. Wikidata records it as a web-based molecular flavor-pairing platform. The platform was built around the idea of answering a single question in two ways at once: what goes with a given ingredient, and why. One answer comes from molecular food science; the other comes from a large corpus of curated recipes. The interesting part, as the platform describes it, is where the two signals agree and where they disagree. The data foundation consists of 857 ingredients, 5,240 flavor compounds, and more than 610,000 curated recipes (612,557 cited on the homepage). The platform reports 298,783 scored pairings and grounds its work in more than 40,000 peer-reviewed papers (44,648 cited on the homepage). Scoring methodology evolved around two signals. The molecular signal compares two ingredients by how much of their volatile vocabulary they share, weighted by how rare each compound is (IDF-weighted Jaccard). The recipe signal uses normalized pointwise mutual information over the recipe corpus to see which ingredients chefs actually pair. Default scoring is 50% molecular plus 50% culinary, with a synergy bonus when both signals agree. A Science mode offers pure molecular scoring (70% compound Jaccard plus 30% aroma-vector cosine similarity). The platform flags novel pairings where the two signals disagree — when two ingredients share many flavor compounds but rarely appear in recipes together. It cites examples such as chocolate and blue cheese, or strawberry and coriander, as combinations that end up on famous tasting menus. Product development expanded from chef-facing tools (Pair Finder, Novel Pairings, Bridge Finder) to educator offerings (classroom plans, academic pricing) and to flavor and product-development teams (JSON API, bulk exports, open methodology, scoped pilots on a customer's own ingredient list). The platform also targets recipe and content platforms, CPG and flavor R&D, and meal kits and grocery. Pricing tiers include a free tier, Pro at $12 per month, Academic at $49 per year, Educator at $49 per year, Classroom at $199 per year, and Enterprise API with team licenses and bulk data. A free weekly newsletter offers a molecular breakdown for cooks, instructors, and flavorists.
Selected for the Georgia Tech Food & Beverage Incubator
Profiled by Hypepotamus
Platform reports 612,557 curated recipes, 857 ingredients, 5,240 flavor compounds, and 44,648 research papers
Platform reports 298,783 scored pairings
Products and positioning
A transparent, paper-grounded molecular flavor-pairing platform for chefs, educators, and flavor R&D teams — not a black-box prediction.
Pair Finder
Best matches for any ingredient, with science or classic scoring.
Novel Pairings
Combinations chefs have not explored yet, flagged where molecular and recipe signals disagree.
Bridge Finder
A tool for finding bridging ingredients between flavors.
Flavor Data API
JSON API with bulk exports and open methodology for flavor and product-development teams.
Science mode
Pure molecular scoring: 70% compound Jaccard plus 30% aroma-vector cosine similarity.
Compound glossary
5,240 flavor compounds with example compounds such as vanillin, eugenol, limonene, linalool, menthol, and maltol.
Recipe corpus
More than 610,000 curated recipes used for recipe co-occurrence scoring.
Weekly molecular newsletter
A free weekly molecular breakdown for cooks, instructors, and flavorists.
Recent events
Reported
Huobi DeFi Labs announces the Global DeFi Alliance[5]
Huobi DeFi Labs announced the launch of the Global DeFi Alliance with founding participants including Huobi, the Maker Foundation, Compound, NEST communities, and dYdX. The alliance is intended to advance DeFi research and development, promote best practices, and facilitate cross-border collaboration.
Reported
ZenGo launches savings mode[6]
ZenGo added a savings mode that lets users set aside crypto assets to earn interest, with assets accessible through DeFi projects including Compound and Figment.
Reported
Gauntlet announces partnerships with Stellar, Celo and Compound[7]
Gauntlet announced partnerships with Stellar, Celo and Compound to help the teams behind the protocols design, launch and scale new financial systems.
Sources7 sources
- 1CompKitchen — Molecular Flavor Pairingcompkitchen.com
- 2How It Works | CompKitchencompkitchen.com
- 3Flavor Data API — CompKitchencompkitchen.com
- 4Compound Kitchen — Wikidatawikidata.org
- 5Huobi DeFi Labs 宣布成立 Global DeFi Allianceprnewswire.com
- 6ZenGo推出储蓄模式techcrunch.com
- 7Gauntlet宣布与Stellar、Celo和Compound建立合作关系finance.yahoo.com
Cite this profile: Cite the canonical profile. /brand-wiki/compound · Editorial policy · How profiles are compiled