Newton's second law, $\vec F=m\vec a$ assumes we apply forces at the COM.
It does not assumes that. In a Newtonian system of particles, each of the external forces acting on the system is assumed to be applied to some individual particle rather than to the center of mass. However, the dynamical behavior of the center of mass of the system of particles obeys
$$M_\text{tot}\, {\ddot{\vec x}}_{\text{com}} = \vec F_\text{ext,tot} = \sum_i \vec F_{\text{ext};i} = \sum_i \sum_j \vec F_{\text{ext};i,j} \tag{1}$$
Working from left to right, the first equality ($M_\text{tot}\, {\ddot{\vec x}}_{\text{com}} = \vec F_\text{ext,tot}$) looks exactly like Newton's second law. It says that the center of mass of the system accelerates exactly the same as would a point mass whose mass is the total mass of the system of particles and is acted up by the net external force acting on the system. This does not however mean you truly do apply the forces at the center of mass. This is an as-if expression. This distinction becomes very important when looking at the rotational behavior of the system of particles.
The next equality ($\vec F_\text{ext,tot} = \sum_i \vec F_{\text{ext};i}$) says the total external force on the system of particles is the sum of the total external force on each of the individual particles. So what is the total external force on an individual particle? That's what the final equality ($\sum_i \vec F_{\text{ext};i} = \sum_i \sum_j \vec F_{\text{ext};i,j}$) tells us: The total external force on an individual particle is the sum of the external forces acting on that specific particle. Forces are additive, vectorially.
Whether all of this is meaningful depends on whether the system boundary is meaningful. Imagine a rock on the Moon and a chunk of ice on Pluto. One could denote these two objects as comprising a system of particles. This is a rather meaningless system of particles. While equation (1) does faithfully describes the time evolution of the center of mass of this system, it's not very meaningful. There are plenty of applications where a partitioning into system versus external does make sense, and it's these applications where equation (1) is very helpful.