Development and implementation of an AI-based demand forecasting and purchasing tool for end-to-end optimisation of cream cheese manufacturing

Co-funded by the European Union
prevai-project-quescrem

PREVAI is a cloud‑based platform that applies artificial‑intelligence models to predict demand with greater accuracy. By bringing together data streams from purchasing, sales and production, it gives food companies—including early adopter Innolact (Quescrem) — a single, real‑time view of what will be needed and when. The result is smarter production planning, leaner inventories and less food waste across the supply chain.

The food industry—especially sectors dealing with perishable products like dairy—faces significant challenges in accurately forecasting demand. Commercial managers must make precise predictions in a global market characterized by shifting and region-specific behaviors. Additionally, the lack of integration between purchasing, production, and sales systems leads to inefficiencies, overproduction, and wasted resources.

Solution

PREVAI is a cloud-based digital solution that connects and analyzes purchasing, sales, and production data. It builds different predictive models for each analysis based on historical behavior and integrates them with market information obtained through APIs and web scraping. This solution enables more accurate demand forecasting, improves decision-making, and optimizes cream cheese production.

What PREVAI delivers
  • Improved production control and planning — precise forecasts support 4‑week, quarterly and even annual scheduling.
  • Higher profitability through tighter stock control — holding the right inventory at the right time frees up capital and prevents write‑offs.
  • Lower operating costs and a smaller carbon footprint — fewer last‑minute corrections mean less waste, energy use and transport.
  • Smarter purchasing — the platform tracks raw‑material price trends, so buyers can time orders for the best value.

The project has achieved several key milestones, including the development of a platform adaptable to other food industry subsectors. It is built on TripleAlpha’s AlphaCloud, a solution already proven in other industries. A planning and sequencing system tailored to order and demand forecasting has been implemented, along with a forecast confirmation flow integrated into the Quescrem customer portal. The initiative has involved close collaboration with the Applied Computing Laboratory at the University of Vigo, PFS Grupo, Seidor, Partnertec, and Planning Enterprise, with commercial support from the consulting firm Naifman. The project has also gained visibility through participation in European-level food industry forums and trade fairs.

Future outlook

The commercialisation of similar platforms for other companies in the agri-food sector is currently underway. Its modular design allows it to be adapted to the specific needs of each company. Furthermore, the platform is expected to continue evolving with new modules and to be deployed across various European regions through the HIGHFIVE project’s partner network.

Testimonials
1/0

From the commercial area, we highly value the incorporation of the demand forecasting tool. It has allowed us to anticipate purchase variations from our main customers more accurately. We used to work with historical and a lot of intuition, but now we have updated data and models that indicate possible peaks in demand weeks in advance. 

This has facilitated both planning with production and negotiation with distributors. There is still room for improvement in some segments, such as new products or one-off promotions, where forecasting is less accurate. But overall, the impact has clearly been positive for our planning and decision-making.

Commercial department of Innolact