The construction of polynomials $R \mapsto R[x]$ gives a functor $P: \mathbf{Ring} \to \mathbf{Ring}$ on the category of possibly noncommutative rings. Choosing a ring $R$ for the moment, there is a nice homomorphism $R \to P(R)$ which embeds in the obvious way, taking $r$ to the constant polynomial $r \cdot 1$. There also seems to be a homomorphism $P(P(R)) \to P(R)$: given a polynomial in $x$ with coefficients in $R[x]$, just do the multiplication and addition to get a polynomial with coefficients in $R$.
This sounds suspiciously like a monad on $\mathbf{Ring}$. I think that the above maps are in fact natural transformations $\eta: 1_{\mathbf{Ring}} \to P$ and $\mu: P^2 \to P$, and that $\mu \circ P \eta = \mu \circ \eta P = 1_P$ and $\mu \circ P \mu = \mu \circ \mu P$.
Is this right? And if so, what have people done with this idea? For example, I see that there are things called polynomial monads, but it's not clear how they might be related.