ESPRESSO, the next generation ESO VLT high-resolution ultra-stable spectrograph, after the successful Preliminary Acceptance Europe held at the integration site of the Observatory of Geneva, has been re-integrated at Paranal and started its commissioning activities at the end of 2017. One critical aspect for ESPRESSO future operations, compared with other instruments currently running at ESO, is the way it will be operated which poses several constraints on its data flow. ESPRESSO has been conceived and developed as a "truly science-grade products generating machine" thanks to its fixed format and long-term stability. In addition to the Data Reduction Software (DRS), a Data Analysis Software (DAS), developed within the standard ESO Data Flow System, will be provided to the users - a novelty for the instruments at Paranal. Moreover, ESPRESSO will be fed either by the light of any of the UTs or by the incoherently combined light of up to four UTs, a feature which required a re-thinking of the current Paranal data handling injection schema. In this paper, after describing the main challenges and peculiarities of the ESPRESSO data flow system listed above, we will present the results of the first commissioning activities and the lessons learned to handle data produced by an instrument with such ambitious scientific requirements.