Argo is a global array of profiling floats that provides temperature (T) and salinity (S) profiles from 2000 m to the surface every ten days with a nominal spatial resolution of 3 degrees. Here we present idealized experiments where the adjoint method is used to synthesize simulated sets of Argo profiles with a general circulation model, over a one-year period, in the North Atlantic. Using a number of drifting profilers consistent with Argo deployment objectives, the simulated array permits one to identify large-scale anomalies in the hydrography and circulation, despite the presence of a simulated eddy noise of large amplitude. Model dynamics provide an objective means to distinguish eddy noise from large-scale oceanic variability, and to infer the absolute velocity field (including abyssal velocities and sea surface height) from sets of Argo profiles of T and S. In particular, our idealized experiments suggest that volume and heat transports can be efficiently constrained by sets of Argo profiles. Increasing the number of Argo floats seems to be an adequate strategy to further reduce errors in circulation estimates.