Folding simulation and box size

GROMACS version:
GROMACS modification: 2023.3
I would like to simulate a folding process of a protein with some disordered regions (in silico modeled protein). But the long N-terminal tail makes the dodecahedron box too big to run the simulation in a rational amount of time.
Are there any valid or standard approaches that can be used in such case (may be simulate protein by parts or change the simulation box for smaller as protein becomes more compact)? Or may be I need some additional tools to be used in order to simplify the task?