AI-based Supply Chain Optimization 

CORTEX aiSUPPLY is an end-to-end integrated AI-based Supply Chain Optimization solution using all available structured and unstructured data. Forecast accuracy and demand variability are some of the top obstacles to achieving retailers’ supply chain goals. AI-based supply chain optimization in retail, for example, enables significant automation, reduced write-offs and waste, capital expenditure, out-of-stock, but also increased freshness, turnover and efficiency. The solution incorporates many influencing factors such as price, individual customer promos, promotion rules, local competitor, competitor price, competitive articles, coupon promotions, TV ads, print/online promotions, vendor/retailer promotions, seasonality, local holidays, day of the week or month, brand, weather and weather forecast.

Replenishment Optimization typically involves machine learning-based demand forecasts where all relevant data analyzed and a probabilistic demand forecast are created for each product and location over a demand period. It considers the replenishment constraints and goals with respect to inventory and deliveries. Automated decisions lead to optimized replenishment orders which results in significantly reduced out-of-stock, waste, and manual intervention and increased availability, but also less items on stock.

For spare parts replenishment, the distribution of spare parts from warehouses at different locations are considered with business decisions made on costs versus service level. A predictive solution can optimize risk of over-capacity and out-of-stock situations. A full risk profile on SKU level allows the optimization of single warehouses and lower costs of storing, transporting and unplanned deliveries.

When price optimisation for retailers are considered, AI-based pricing in retail enables significant automation, increased market share, turnover, raw profit, customers and new customers and a reduction in returns, complaints and rests at end of season. Older pricing strategies include the following each with their issues:

  • Not all prices can be increased by 1% because of competition. On the other hand, some prices can be increased much more. But which prices can be increased and by how much?
  • Originating as a loss-prevention strategy, odd pricing (such as “9.99”) has been declared a “psychological pricing technique”. Not statistical evidence of its success.
  • A cost-plus pricing strategy typically economically inconsistent by ignoring customers and consumers.

AI-based price optimization starts with measuring of price elasticities by identifying relevant prediction features, calculating relevant price elasticities of demand per store and per product, and learning the continuously changing price-demand relations using data. Machine learning–based price optimization involves the application of strategic price rules, considering location- individual stock levels, storage costs, etc., and applying margin and revenue expectations according strategic goals. The results are an optimized price per product and channel (shop, online), highest level of automation, and price calculation according to company strategy.


The future of AI in Supply Chain always starts from customer demand and is completely data-driven and scientific with demand predictions at the finest grain. It also involves aggregating up the supply chain, planning higher levels, breaking down silos and holistic company optimization, the merge of the supply chain and pricing (algorithmic retailing), and data exchange even across supplier – retailer boundaries.

Relevant industries