An accreting dwarf star orbiting the S-type giant star pi1 Gru

Aims. We aim to characterize the properties of the inner companion of the S-type AGB star pi1 Gru and to identify plausible future evolutionary scenarios for this triple system. Methods. We observed pi1 Gru with ALMA and VLT/SPHERE. In addition, we collected archival photometry data and used the Hipparcos-Gaia proper motion anomaly. We derive the best orbital parameters from Bayesian inference. Results. In June-July 2019, the inner companion, pi1 Gru C, was located at 37.4 +/- 2.0 mas from the primary (a projected separation of 6.05 +/- 0.55 au at 161.7 +/- 11.7 pc). The best orbital solution yields a companion mass of 0.86 (+0.22/-0.20) Msun (using the derived mass of the primary) and a semi-major axis of 7.05(+0.54/-0.57) au, corresponding to an orbital period of 11.0 (+1.7/-1.5) yr. The preferred solution is an elliptical orbit with eccentricity e = 0.35(+0.18/-0.17), although a circular orbit cannot be fully excluded. The close companion could be either a K1V (F9.5V to K7V) star or a white dwarf (WD). Ultraviolet and millimeter continuum photometry are consistent with the presence of an accretion disk around the close companion. The ultraviolet emission may originate from hot spots in an overall cooler disk, or from a hot disk if the companion is a WD. Conclusions. Although the close companion and the AGB star are interacting and an accretion disk is observed around the companion, the mass-accretion rate is too low to trigger a Type Ia supernova, but it could produce novae every ~900 yr. Short-wavelength, spatially resolved observations are required to further constrain the nature of the C companion. Searches for close-in companions similar to this system will improve our understanding of the physics of mass and angular momentum transfer, as well as orbital evolution during late evolutionary stages.