ItalyConsumer goods2026 – ongoingIntercos
In-house predictive AI models for cosmetic formulation testing, cost forecasting and raw-material document review
Intercos built its own AI models to predict lab-test results for new cosmetic formulas
Problem
Developing a new cosmetic product means repeated lab tests, and some formulations fail. Technical and regulatory documents for each raw material also take time to review.
Approach
Intercos built AI models in house, with Accenture as a named partner, over about a year. One model predicts the outcome of certain lab tests, including the chance of failure, to shorten development. A second estimates the full industrial cost of a product by comparing it with similar past products. A third reviews technical and regulatory documents for raw materials and was already built. Tools to find suitable packaging were in testing, and the lab-test model was due to go live within weeks.
Results
No published figures
Takeaway
Intercos aimed the models at three routine bottlenecks in product development: failed tests, cost estimates and document checks. It kept its data and knowledge inside the company.
Categories
- Industry
- Consumer goods
- Business function
- IT and R&DProduction and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Partner co-development
- Evidence
- Press reporting
Sources
Source summaries
An interview with Intercos's technology chief describes four AI strands: lab-test prediction for new formulations, industrial cost forecasting, review of raw-material documents and packaging search. The document review was already built, the lab-test tool was due to go live within weeks and packaging search was in testing. Accenture is named as a partner. The article gives no figures.
Full details are in the original sources.
