Formulation design cycle reduced by 70% and head OEM manufacturers competed for layout

Formulation design cycle reduced by 70%

Data source: PCHI 2026 Hangzhou Exhibition, Wah Hee-Bian, Head contract manufacturing Factory, Technical Guide to the Central Inspectorate

From 18 to 20 March, PCHI 2026 was held at the Hangzhou Congress Centre, with AI supporting formulation R&D and synthetic biological materials as the highlight of the exhibition. A number of head OEM manufacturers present the AI formulation system in use, with measured data: formulation design cycle reduced from two weeks to three days, and stability projected a hit rate of 78 per cent (approximately 65 per cent manual experience). This is a sign that the make-up industry is undergoing profound changes from "experience driven" to "data driven".

I. Three core AI applications in formulation R&D

Application 1: smart material referral system

Traditional mode pain point:

• The formulation division relies on personal experience and has thousands of raw materials that are difficult to fully cover.
• High test error cost: one formulation often requires 5-10 rounds of small samples
• Manual combination optimization: 3 ingredient combinations need to test 7 groups, 10 ingredient combinations need to test 1023 groups

AI Solutions:

• Enter target efficacy (e. g. "anti-wrinkle + brightening + sensitive skin applies") to generate system seconds for candidate formulation
• Ingredient based on million-degree formulation database, smart matching synergies
• Automatically circumvent ingredient conflicts (e. g. co-use stability)

Results measured:

Applying 2: stability prediction model

Traditional mode pain point:

• Stability testing Time-consuming: Accelerating ageing (40°C/75% RR) takes 3 months, real ageing takes 6-12 months
• Artificial empirical judgement: Relying on the "intuitive" of a senior formulation division, the new formulation division is unmanageable
• The cost of failure is high: mass production was found before the problem of stability, meaning that the R & D cycle is zero for months

AI Solutions:

• Based on historical formulation data training model projected for 3 months, 6 months, 12 months stability
• Identification of potential problems: pH drift, emulsification layer, colour variation, crystal
• Recommended optimisation programme: adjustment of type of denser agent, replacement of preservative system

Results measured:

According to data displayed by Wah Hee-Wu, its AI stability prediction system:

• Hit rate 78 per cent: 78 per cent of formulation projected to be unstable did show problems in subsequent measurements
• 15% false positive: the percentage of miscalculated unstable is within 15%
• 85% recall rate: 85% real problem identified

Application of 3: efficacy ingredient That's right.

Traditional mode pain point:

• Efficacy ingredient (e.g. retinol, niacinamide) concentration needs to balance effectiveness with safety
• Single concentration difficult to meet different skin needs (sensitive skin vs. resistance muscles)
• Ingredient Synergy is difficult to quantify (e.g. VC+VE synergies)

AI Solutions:

• Based on clinical databases, create a "concentration - efficacy - safety" 3D model
• Enter skin type (sensitive/dry/oily) and recommend best concentration range
• Simulate ingredient synergy: calculate efficiency factor for multiple ingredient combinations

Actual cases:

Anti-wrinkle, developed by a head OEM manufacturer, optimized through the AI system:

• Old formulation: retinol 0.1% + botanical 2%
• AI Optimization: retinol 0.15% + botanical 3% + ergothioneine 0.5%
• Result: Wrinkle rating of VISIA increased by 43 per cent (28 per cent as originally planned) and irritation potential test rating down from 3 to 1

II. Status of the AI layout of head OEM manufacturers

Wah Hee-Bian: full chain AI enablement

Research and development end:

• AI Raw Material Recommended System: Integration of 8,000+ Raw Material Data Library
• Stability prediction model: cover 2000+ history formulation
• Efficacy simulation platform: data with third-party testing institutions Fight! All

Production end:

• AI process optimization: fermentation parameters are automatically adjusted to increase production rate by 15%
• Smart quality inspection: Visual recognition system detection of visual defects, leakage rate < 0.5%

Result: Reduction in the research and development cycle for new products from an average of 8 months to 3 months in 2025.

Cosmic Poet (South Korea contract manufacturing plant giant): formulation AI

Deployment:

• Online at the Shanghai Research and Development Centre, AI formulation system, 2025
• Category

Core functions:

• Smart formulation generation: supports the efficacy type, basic, and makeup model with three main formulation modes
• Cost optimization module: automatic search for the highest value-for-money combination of raw materials with guaranteed efficacy

Outcome: 80 per cent of the new product formulation was generated by the AI system from the second half of 2025.

Fu Fu Xian Electrician (perfect diary parent company): AIX Raw Materials Innovation

AI-driven feedstock development:

• Work with the Great Chinese Genome to create dermal microecological data Library
• AI screening of probiotics fermentation products. Three new ingredient with anti-inflammatory potential were found

AI supports formulation customization:

• Formulation
• "AI z liquid foundation " , introduced in 2025, supports the matching of 200+ color

III. Technical architecture of the AI formulation system

Data layer: three data sources

Algorithm: Three core algorithms

1. Graphical nerve network (GNN)

• Use: Synergy/resistence between ingredient
• Enter: ingredient molecular structure
• Output: ingredient compatibility rating

Random Forest

• Use: stability projection
• Input: formulation ingredient +pH+preservative system
• Output: 3/6/12 months stability probability

3. Enhanced Learning

• Purpose: Multi-target optimization (efficacy + cost+stability)
• Input: formulation target (efficacy target, cost cap)
• Output: Best formulation scheme

IV. Implications and opportunities for OEM industry

Impact I: transformation of the formulation division

From Master of Experience to "Ai."

• Primary formulation Division: Learn to use AI tools to quickly generate the first draft of formulation
• Senior formulation Division: Focus on areas where AI cannot be replaced (sensitization, interpretation of regulations)
• New industry standard: not using AI formulation division, possibly phased out in three years

Impact II: Reduction in the revolutionary nature of the new product development cycle

Traditional vs AI aid cycle:

Commercial value:

• Increase in the annual volume of new products from 2 to 3 to 4 to 6
• Market window: Your product was on the market while the competition was still under development.

Impact III: Increased head effect

Contract manufacturing plant advantage with AI system:

• Low R & D costs to provide more competitive quotation
• Brand owner is willing to cooperate.
• Data accumulate and AI models are becoming more accurate (Marta effects)

Challenges for small and medium contract manufacturing plants:

• High cost of AI system development (approximately $5-8 million)
• Technical team required (data engineer + AI algorithm)
• It's hard to develop in the short term.

Breaking path:

• Joining the AI formulation SaaS platform (projected to be launched by multiple manufacturers by 2026)
• In cooperation with the head contract manufacturing plant, use its AI system (pay-for mode)

V. Three points for QuickOEM users

Recommendation 1: Assessment of the AI capabilities of the contract manufacturing plant

Highlights of the mission:

• Is there an AI formulation system (non-concept, already produced)
• Actual effects data for AI system (stability forecast hit rate, formulation design cycle)
• Whether brand owner is allowed for trial or observation

List of issues:

• "How long has your AI system been online?"
• "What's the expected hit rate of stability from AI?
• "Can you provide the formulation case generated by the AI system and the results?"

Recommendation 2: Cooperation with AI pilot contract manufacturing plant

Priority selection:

• Wah Hee-Wah: Synthetic + AI double-wheel drive, technically powerful
• Cosmic: International background, formulation AI mature
• Local head OEM: already online AI system contract manufacturing plant

Cooperation modalities:

• Standard cooperation: factory AI system to generate formulation
• Customization development: training for the exclusive AI model based on brand needs

Recommendation 3: Brand-owned AI capabilities

Short term (within six months):

• Collection of product research and development data (formulation, stability, efficacy tests)
• Create brand formulation database

Medium term (1 year):

• Countering third-party AI formulation platform (e.g. AI tool provided by feedstock provider)
• Training of formulation division on the use of AI tools

Long term (1 - 2 years):

• Consider building or jointly developing a proprietary AI formulation system
• IA capacity as a brand differential selling point

Industry outlook

The Technical Guide for the Purposes of Use of Cosmetic Materials (Preliminary) issued by the Central Prosecutor's Office in December 2025 clearly states: "Encourage enterprises to optimize the design and increase the efficiency of research and development using new technologies such as artificial intelligence". This marks a formal recognition of AI ' s value in cosmetics development.

By 2028:

• Head OEM manufacturer AIS coverage is 90%.
• AI supports formulation into industry standard processes
• The formulation division, which will not use AI, will face elimination.

If the OEM enterprise is able to seize this historic opportunity, it will create an absolute advantage in efficiency, cost, innovation capacity, from "contract manufacturing plant" to "formulation innovation service provider".

View QuickOEM formulation R&D services

Data sources

• PCHI 2026 Hangzhou Exhibition site research
• "Al enablement Cosmetics Development" White paper
• Cosmier Poet Network technical files
• Technical Guide to the Purpose of Use of Cosmetic Materials (Preliminary) by the Central Prosecutor ' s Office

Other Organiser

This article is published by the QuickOEM Make-up OEM Smart Matching Platform. More makeup OEM knowledge, cosmetics contract manufacturing, skin protection private label, make-up customizationontent are available at the Knowledge Centre. If you need advice on OEM contract manufacturing, please contact 1814805760 or submit your request through an online request for quotation.

Keywords: PCHI 2026 Fair, Synthetic Organisms, Perfect Diaries, AAI Cosmetics Development, OEM, contract manufacturing, contract manufacturing Factory, Make-up AI Application, Wahe-Bio, OEM Digital Transformation, emulsification, smart formulation System, ViC, niacinamide, Aol, Information on the Make-up Industry, Trends in the Cosmetic Industry, Cosmetic Market Dynamics, Cosmetic Policy Regulations

QuickOEM is a QuickOEM platform for vertical digitization of cosmetics services (known as QuickOEM) focusing on OEM, cosmetics contract manufacturing, skins private label, cosmetics z, brand-wide, ODM contract manufacturing services. AI smart matches the 500+ certification contract manufacturing plant, providing a one-stop package from formulation R&D to regulatory compliance (EU / US FDA) covering skin, makeup, sheet mask, serum, eye patch, category, category, etc.

Core functions

• Makeup OEM needs to be published to match factory AI intelligence
• Cosmetics contract manufacturing / skin protection private label customization ODM service
• Formulation R&D management and multi-version tracking
• Ingredient Compliance testing (regulatory compliance (EU / US FDA)/EU 1223/2009/FDA)
• Regulatory compliance (EU / US FDA) Cosmetics regulatory filing fully tracked
• Supply chain
• Advertising Law Compliance Testing
• Z / eye patch / serum / z all category

Please enable JavaScript to use the full functionality.

Login immediately.

WhatsApp: +1 (646) 509-6134