Preparation of thermal and ground states of many-body systems is a central challenge for quantum processors, needed e.g. as the starting point for many quantum physics experiments or for quantum chemistry applications. In this talk, I will discuss recent work on efficient state preparation using engineered system-bath physics. First, I will overview results from the Google experiment [Science 383 6689 2024], where a version of the dissipative algorithm was used to prepare low-energy states of quantum magnetic systems. I will then explain how to modify the algorithm to accurately prepare thermal and ground states [arXiv:2506.21318 & PRX Quantum 6, 010361 2025]: the resulting algorithm is suitable for near-term quantum devices, and exhibits partial robustness to noise owing to the dissipative nature. We demonstrate its efficiency numerically for ground states of 1d chain and ladder systems, and thermal states of the 2D quantum Ising model, providing perturbative arguments for its more general validity.