We developed innovative experimental tools and advanced disease models, enabling major discoveries into the mechanisms by which pancreatic cancer evolves and spreads. We identified and functionally characterized key drivers and modulators of metastasis and demonstrated that metastatic capacity is often determined very early, at the stage of the primary tumor. In most cases, metastatic tumors were genetically highly similar to their primary tumors, challenging the traditional view that metastasis develops gradually through the accumulation of late mutations. Among the identified drivers, amplification of mutated KRAS emerged as a central metastasis switch. While mutation of the KRAS gene represents the initiating event in pancreatic cancer development, we discovered that a subsequent increase in its dosage activates a program that enables cancer cells to disseminate. Crucially, we uncovered why this amplification arises so early: elevated KRAS activity is required for cellular dedifferentiation, the first step in malignant transformation. Thus, the very event that drives early tumor initiation — amplification of mutated KRAS — simultaneously equips cells with a metastatic program, providing a unifying mechanistic explanation for the early acquisition of metastatic competence in pancreatic cancer. The genetic technologies, disease models, and methods developed in PACA-MET have been widely disseminated and are now used by researchers worldwide. Their broad applicability accelerates progress across disciplines.