A plan you can explore.
Explore the result of a real multi-agent planning run: regional distribution, loading layouts, and the evidence behind the selected strategy.
Synthetic products · Precomputed results · No live model calls
Loading interactive case…
What this case demonstrates
Seeded synthetic products, inspired by the original test dimension ranges, were processed by the existing Python API and independent worker. Matched product pairs exercise side-by-side and multi-row loading. Four strategy agents generated and evaluated candidates; the selected plan passed independent physical checks. LLM assistance was disabled.
Read the numbers in context
All product records in MP-02 are synthetic. The test covers 18 provinces with deliberately varied dimensions and demand, targeting four to five wagons per region. This is an illustrative case, not a real order or a maximum-utilisation benchmark. Counts refer to planned wagons, not confirmed departures; displayed geography is province-level only.