KosmetikOn AI Infrastructure
kAI — Purpose-Built AI for Your Industry
Generic AI knows a little about everything. kAI knows everything about cosmetics, fragrance, and haute cuisine. Built from the ground up on the largest proprietary specialized datasets in these industries, kAI is the intelligence layer that powers every module of the Labify® platform — delivering domain-specific AI that general-purpose models cannot replicate.
kAI is KosmetikOn's proprietary AI layer — a purpose-built intelligence engine trained on the largest proprietary specialized datasets in the cosmetics, fragrance, and culinary science industries, including over 100,000 raw material profiles with structured data on INCI nomenclature, function, safety, regulatory status, recommended concentrations, and supplier provenance.
Developed and led by CTO Pere Adell Parramon — an EPFL engineer (BSc Communication Systems 2017, MFE Financial Engineering 2019) and author of "Forecasting Beta Using Machine Learning and Equity Sentiment Variables" published in Machine Learning for Asset Management (Wiley, 2020) — kAI was founded as a dedicated AI research initiative in 2019 and has since become the intelligence backbone of every Labify® product and La Dalle.
Unlike generic large language models applied as an afterthought to cosmetics workflows, kAI performs formulation-relevant tasks that require structured, current domain knowledge — including Margin of Safety (MOS) calculations, IFRA limit checks, allergen cross-referencing, ingredient suggestions based on regulatory constraints, and continuous regulatory compliance monitoring across EU, US, and ASEAN jurisdictions.
Built by AI Researchers, Not Consultants
kAI is the product of a dedicated AI research team with published credentials in machine learning — applied exclusively to the domains where KosmetikOn operates.
Pere Adell Parramon
CTO & kAI Lead Engineer, KosmetikOn
- Education: EPFL (École Polytechnique Fédérale de Lausanne) — BSc Communication Systems (2017), MFE Financial Engineering (2019)
- Published research: "Forecasting Beta Using Machine Learning and Equity Sentiment Variables" — chapter in Machine Learning for Asset Management, edited by Emmanuel Jurczenko, published by John Wiley & Sons (2020)
- Role at KosmetikOn: Founded and leads the kAI research team since 2019. Responsible for the architecture, training, and continuous development of kAI's specialized AI agents and proprietary datasets
The kAI research team was founded in 2019 with a single mission: build AI that understands the cosmetics, fragrance, and culinary industries at a level of specificity that no general-purpose model can approach.
The Largest Specialized Datasets in the Industry
kAI's domain specificity comes from the scale and depth of its proprietary data — continuously maintained and updated, not a static training snapshot.
Raw Material Profiles
Over 100,000 raw material profiles with structured data including INCI nomenclature, chemical function, safety profiles, regulatory status across multiple jurisdictions, recommended use concentrations, known interactions, and supplier provenance information.
Toxicology & Safety
Comprehensive toxicological databases covering ingredient safety assessments, allergen identification, cross-reactivity analysis, and Margin of Safety (MOS) reference data across multiple consumer age ranges — enabling automated safety calculations directly from formulation data.
Regulatory Frameworks
Continuously updated regulatory databases across multiple jurisdictions: EU Cosmetics Regulation 1223/2009, REACH (Registration, Evaluation, Authorisation and Restriction of Chemicals), IFRA standards for fragrance, US FDA frameworks, ASEAN cosmetics directives, and EU Food Information Regulation (FIR) for the culinary vertical.
Market & Trend Data
Predictive trend and sentiment analysis data covering cosmetics, fragrance, and culinary markets — enabling formulators and R&D teams to align product development with emerging market demands and consumer preferences.
Olfactive Profiles
Fragrance-specific datasets covering ingredient olfactive profiles, olfactive family classification, IFRA amendment history, volatility profiling, and evaporation modelling — purpose-built for perfumers and fragrance formulators working within Labify® Nez.
Culinary Science
Culinary-specific datasets covering ingredient flavour profiles, food science literature, allergen and nutritional data, seasonal availability, provenance information, and culinary trend data — powering the AI layer within La Dalle for haute cuisine R&D teams.
What kAI Does — Across Every Module
kAI provides specialized AI agents accessible to every platform user with unlimited requests. Throughput and response speed may vary during periods of peak traffic.
Regulatory Compliance Monitoring
Continuous automated checks against current regulatory frameworks — EU Cosmetics Regulation 1223/2009, REACH, IFRA standards, and others. kAI generates alerts when regulatory changes affect existing formulations and validates new formulations before production.
Margin of Safety (MOS) Calculation
Automatic MOS calculation across multiple consumer age ranges, directly from formulation data — a critical regulatory requirement for cosmetic product safety assessment that kAI performs with structured, current toxicological data.
IFRA Limit Checks
Automated compliance checking against current IFRA (International Fragrance Association) standards for fragrance formulations — including limit calculations per product category, allergen declaration management, and reformulation guidance when standards change.
Allergen Cross-Referencing
Identification of allergens and cross-reactivity risks across ingredients in a formulation — covering cosmetics allergens (EU Cosmetics Regulation), fragrance allergens (IFRA/EU labelling), and food allergens (EU Food Information Regulation) depending on the vertical.
Ingredient Suggestions
AI-powered ingredient recommendations based on target claims, regulatory constraints, performance gaps, or reformulation needs — connecting formulators directly to the 100,000+ raw material database with suggestions that are compliant, available, and formulation-relevant.
Predictive Trend & Sentiment Analysis
Market intelligence and trend analysis covering cosmetics, fragrance, and culinary categories — helping R&D teams align product development with emerging demands and stay ahead of market shifts.
Sustainability Scoring
Sustainability assessment for formulations and ingredients — evaluating environmental impact, sourcing practices, and ingredient provenance to support increasingly important sustainability commitments.
Why Domain-Specific AI, Not a Generic LLM
General-purpose large language models are powerful tools for general knowledge. They are not reliable tools for formulation-level decisions in regulated industries. Here is why kAI exists.
Structured Data vs. Probabilistic Text
A generic LLM generates text based on statistical patterns. kAI operates on structured, verified datasets — 100,000+ raw material profiles with specific INCI names, CAS numbers, regulatory statuses, concentration limits, and safety data. When kAI calculates an MOS value or checks an IFRA limit, it is querying structured data, not generating a probable answer.
Current vs. Frozen Knowledge
Generic LLMs have knowledge cutoff dates. Regulatory frameworks in cosmetics, fragrance, and food change frequently — IFRA amendments, EU regulatory updates, REACH substance evaluations. kAI's datasets are continuously maintained and updated by KosmetikOn's domain experts. The AI's knowledge is always current because the data it references is always current.
Formulation-Relevant vs. Conversationally Plausible
Ask a generic LLM about cosmetic ingredients and you will get a plausible-sounding paragraph. Ask kAI and you will get a formulation-relevant answer with specific concentration ranges, regulatory restrictions by jurisdiction, known interactions with other ingredients, and supplier availability — because kAI was built by domain experts to serve domain experts.
Integrated vs. Isolated
kAI is not a chatbot bolted onto a software platform. It is the intelligence layer woven through every module — PLM, CRM, SRM, ERP, Accounting, and HR. kAI's suggestions, validations, and analyses operate on the actual formulation data, manufacturing records, and supplier information already in the platform. A generic LLM has no access to this context.
kAI Across Three Industry Verticals
The same AI architecture, trained on industry-specific datasets — kAI adapts to the regulatory, formulation, and workflow requirements of each vertical it serves.
kAI for Cosmetics — Labify® Beauté
Regulatory compliance against EU Cosmetics Regulation 1223/2009 and REACH. Automatic MOS calculations across age ranges. INCI list validation. Allergen identification and cross-reactivity analysis. Ingredient suggestions based on target claims and regulatory constraints. Sustainability scoring for formulations. Predictive trend analysis for cosmetics markets.
Explore Labify® Beauté →kAI for Fragrance — Labify® Nez
IFRA compliance automation against current standards. Allergen substitution recommendations. Olfactive family classification and benchmarking. AI-assisted accord suggestions and reformulation support. Volatility profiling and evaporation modelling. Regulatory change tracking specific to fragrance ingredients. Market trend analysis for fragrance categories.
Explore Labify® Nez →kAI for Haute Cuisine — La Dalle
Ingredient substitution suggestions and flavour pairing recommendations. Allergen-safe reformulations aligned with EU Food Information Regulation (FIR). Nutritional analysis support. Provenance and sustainability scoring for ingredients. AI-assisted creative development that supports the chef's vision without replacing it. Culinary trend intelligence.
Explore La Dalle →Frequently Asked Questions About kAI
What is kAI and how does it differ from generic AI tools like ChatGPT?
kAI is KosmetikOn's proprietary AI layer, purpose-built for the cosmetics, fragrance, and haute cuisine industries. Unlike generic large language models such as ChatGPT, kAI is trained on the largest proprietary specialized datasets in these fields — including 100,000+ raw material profiles, toxicology databases, regulatory frameworks (EU, US, ASEAN), and ingredient safety assessments. This domain specificity allows kAI to perform formulation-relevant tasks such as Margin of Safety (MOS) calculations, IFRA limit checks, and allergen cross-referencing that require structured, current domain knowledge — capabilities beyond what a general-purpose LLM can reliably provide.
Who developed kAI and what are the team's credentials?
kAI was developed and is led by Pere Adell Parramon, CTO of KosmetikOn and EPFL engineer (BSc Communication Systems 2017, MFE Financial Engineering 2019). He is the author of "Forecasting Beta Using Machine Learning and Equity Sentiment Variables," a chapter published in Machine Learning for Asset Management, edited by Emmanuel Jurczenko (John Wiley & Sons, 2020). The kAI research team was founded in 2019 and has built the AI layer from the ground up with deep domain expertise in cosmetics, fragrance, and regulatory science.
How large is the kAI dataset and what does it cover?
kAI is trained on the largest proprietary specialized datasets in cosmetics, fragrance, and culinary science. This includes over 100,000 raw material profiles with structured data on INCI nomenclature, function, safety profiles, regulatory status across multiple jurisdictions, recommended concentrations, and supplier provenance. The datasets also cover toxicology, regulatory frameworks across EU, US, and ASEAN jurisdictions, ingredient safety assessments, market trend data, olfactive profiles (for fragrance), and culinary science data. All datasets are continuously maintained and updated — not a static training snapshot.
Are there limits on how many requests I can make to kAI?
kAI provides specialized AI agents with unlimited requests for all platform users. There is no cap on the number of queries or interactions. Throughput and response speed may vary during periods of peak traffic, but every user has full access to all kAI capabilities without usage-based restrictions.
What specific tasks can kAI perform for cosmetics formulation?
For cosmetics formulation, kAI can: suggest ingredient additions based on target claims, regulatory constraints, or performance gaps; perform automatic Margin of Safety (MOS) calculations across multiple consumer age ranges; monitor regulatory compliance against EU Cosmetics Regulation 1223/2009 and REACH; identify allergens and cross-reactivity risks; generate sustainability scoring for formulations; provide predictive trend and sentiment analysis; and deliver real-time regulatory change alerts. All of these capabilities operate on the actual formulation data within Labify® Beauté — not on a disconnected prompt-response interface.
Does kAI support IFRA compliance for fragrance formulation?
Yes. kAI includes specialized AI agents trained on fragrance-specific datasets including IFRA (International Fragrance Association) compliance standards, ingredient olfactive profiles, IFRA amendment history, and regulatory change tracking. kAI can automate IFRA limit checks, suggest allergen substitutions, assist with accord development, and provide reformulation support when IFRA standards change — all from within the Labify® Nez interface.
How does kAI help with regulatory compliance across multiple jurisdictions?
kAI continuously monitors regulatory frameworks across multiple jurisdictions including the European Union (EU Cosmetics Regulation 1223/2009, REACH), the United States, and ASEAN countries. It provides automated compliance checks against current regulations, generates alerts when regulatory changes affect existing formulations, and validates new formulations against applicable frameworks before they enter production. The regulatory datasets are continuously maintained and updated by KosmetikOn's domain experts.
Can kAI be used for haute cuisine and culinary R&D?
Yes. Through La Dalle — KosmetikOn's platform for haute cuisine R&D — kAI provides AI agents trained on culinary-specific datasets including ingredient flavour profiles, food science literature, allergen and nutritional data, and culinary trend information. kAI assists chefs with ingredient substitution suggestions, flavour pairing recommendations, allergen-safe reformulations, and provenance and sustainability scoring for ingredients — supporting creative development without replacing the chef's creative vision.
See kAI in Action
Book a 30-minute demo and experience how kAI's purpose-built intelligence layer transforms cosmetics, fragrance, and culinary R&D workflows — with domain expertise that no generic AI can match.