What challenges did FM Logistic face before AI implementation?

Reduce unnecessary travel in warehouse routing for ride-on electric trucks. Improve coordination across many picking locations and operators without adding headcount. Get better performance from an already optimized real-time routing baseline.

What was the impact of the AI solution?

Routing efficiency improved by 10.4% over the previous best solution. Warehouse travel is reduced by 15,000+ kilometers per year at full operational scale. FM Logistic can handle larger order volumes with the same team and equipment without adding headcount or expanding the fleet.

Source

GCPProductionEvidence: Medium50/100

FM Logistic: AlphaEvolve + Gemini evolutionary coding agent to optimize warehouse routing

Use case typePlanning automationUpdated Mar 27, 2026

FM Logistic, a global logistics provider operating warehouse operations in Poland, used Google Cloud's AlphaEvolve and Gemini models to improve traveling-salesman-style routing for ride-on electric trucks in a large fulfillment facility. The team seeded AlphaEvolve with FM Logistic's existing routing algorithm and used Gemini to generate and refine code variants, then evaluated candidates with a custom scoring function over 60 representative tours to minimize average travel distance while penalizing operational failures such as capacity, order, FIFO, and computation-time violations. The improved routing logic was deployed in production after the Poland pilot and is intended to support higher order volumes with the same team and equipment.

Organization: FM Logistic
Industry: Logistics
Location: Poland
Published: March 2026

Reported outcomes

15,000 kilometers per year

warehouse travel reductionAdoption & scale

+10.4%routing efficiency improvement

Strategic outcomes

Scale & capacityEnabled higher order volumes with existing resourcesCost efficiencyImproved fulfillment speed and working conditions

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Routing efficiency improvement: 10.4% increase

Google Cloud BlogMar 27, 2026Blog postExplicit claimMedium evidence strength

10.4% improvement in routing efficiency over the previous best solution.

Normalized claim

Warehouse travel reduction: 15,000 kilometers per year decrease

Google Cloud BlogMar 27, 2026Blog postExplicit claimMedium evidence strength

15,000+ fewer kilometers of warehouse travel per year at full operational scale.

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: FM Logistic
Provider: GCP
Maturity: Production
Linked source: Google Cloud Blog

The team seeded AlphaEvolve with FM Logistic's existing routing algorithm and used Gemini to generate and refine code variants, then evaluated candidates with a custom scoring function over 60 representative tours to minimize average travel distance while penalizing operational failures such as capacity, order, FIFO, and computation-time violations

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 2 of 2

1Planning automation
2Workflow automation

Reduce unnecessary travel in warehouse routing for ride-on electric trucks.
Improve coordination across many picking locations and operators without adding headcount.
Get better performance from an already optimized real-time routing baseline.

Seeded AlphaEvolve with FM Logistic's existing routing algorithm.
Used Gemini models to generate and refine algorithm variants.
Built a custom evaluation function over 60 representative tours to test thousands of candidate algorithms under real-world constraints.
Applied penalties for operational failures such as exceeding forklift capacity, missing pending orders, duplicating boxes, violating FIFO priority, or exceeding real-time computation limits.
Deployed the winning routing logic in production after the Poland pilot.

Technologies

AlphaEvolve Gemini

Routing efficiency improved by 10.4% over the previous best solution.
Warehouse travel is reduced by 15,000+ kilometers per year at full operational scale.
FM Logistic can handle larger order volumes with the same team and equipment without adding headcount or expanding the fleet.

Architecture

FM Logistic seeded AlphaEvolve with its existing routing algorithm. AlphaEvolve, using Gemini models, generated and iterated on code variants. FM Logistic designed a custom evaluation function over 60 representative tours and scored candidate algorithms against distance minimization and operational constraints. The improved routing logic was then deployed in production after the Poland pilot.

Sources & evidence1

Evidence: Medium50/100Evidence strength