The solution combined autonomous algorithm search with a practical route from pilot evaluation to production use.

WorkedDid not

Routing logic

AI-generated logic improved efficiency¹

Previous cost-prioritized logic was the baseline¹

Algorithm development

AlphaEvolve generated and refined algorithms¹

Fixed step-by-step decisions alone did not capture the gain¹

Operational outcome

Less travel, faster fulfillment, less fleet wear¹

Existing optimization left further improvement available¹

Deployment

Poland pilot moved into production¹

Pilot-only use was not the final state¹

AlphaEvolve changed the development mechanism. Instead of asking people to manually specify every new routing rule, the coding agent autonomously generated and refined routing algorithms.¹ The approach still had a clear operational test: whether the resulting logic improved routing efficiency and reduced travel.

The reported result was a 10.4% improvement in routing efficiency and over 15,000 kilometres less travel per year at full scale.¹ The Poland pilot’s move into production shows how the generated policy crossed from experimentation into warehouse operations.¹ The case therefore pairs algorithmic novelty with a concrete deployment path.

AI-generated routing logic reduces warehouse travel and boosts routing efficiency