BASF Agricultural Solutions Tests AlphaEvolve on Google Cloud to Build a Data-Driven Supply Chain Digital Twin

BASF Agricultural Solutions is testing AlphaEvolve, an evolutionary coding agent on Google Cloud, to build a digital twin of its crop protection supply chain, according to a May 2026 Google Cloud blog post co-written by BASF and Google. The network has more than 5,000 distinct value chains and 180 production sites, it can take up to two years to turn active ingredients into finished products, and a single end product can need a bill of materials more than 30 levels deep. Planners make thousands of local decisions every day about what to produce, when and how much safety stock to hold, but cannot easily see how a local decision affects the rest of the network, which can tie up working capital and inventory or cause production imbalances. Working with Google Cloud and prognostica GmbH, BASF gave AlphaEvolve a seed program with standard planning logic that translated demand forecasts into production schedules, along with three years of historical data on inventory levels, market demand and actual production. AlphaEvolve generated and mutated variations of the code, and an evaluation function scored each version on how closely its simulated inventory levels and production decisions matched BASF's recorded history. The aim was decision support for planners, not replacing them. In initial tests, the latest runs delivered more than 80% relative improvement in accuracy over the seed model, and the team expects further adjustments to raise accuracy to a level that would make the model actionable for operational use. For the tested product value chain, the evolved, human-readable algorithm discovered rules that explain observed production and inventory: consolidating production amounts to optimise plant time, dynamic safety stocks for volatile and seasonal demand, and coordination across production tiers. "We had several attempts to build a digital twin for our complex supply network using deterministic models, and all of them failed," said Dr. Goetz Krabbe, vice president for global supply chain at BASF. He said AlphaEvolve lets the team map the network from system data while understanding and copying the human decisions that drive daily operations. BASF's goal is to extend the approach to a digital twin of its entire global production network for simulation, decision support, scenario forecasting and optimisation.