Given a monad$t \colon a \to a$ in a 2-category$K$, the Kleisli object$a_t$ of $t$ is, if it exists, the universal right $t$-module or $t$-opalgebra. Equivalently, $a_t$represents the functor $RMod(-,t)$ that takes an object $x$ of $K$ to the category of right $t$-modules $a \to x$.

This means that there is a morphism $f_t \colon a \to a_t$ and a 2-cell $\lambda \colon f_t t \Rightarrow f_t$ that induce an isomorphism$K(a_t,x) \cong RMod(x,t)$: given a right $t$-module $r \colon a \to x, \alpha \colon r t \to r$, there is a unique morphism $a_t \to x$ whose composite with $f_t$ (repsectively $\lambda$) is equal to $r$ (resp. $\alpha$).

For a monad $T \colon A ⇸ A$ in the bicategory Prof of profunctors, its Kleisli object consists of a category $A_T$ equipped with a bijective-on-objects functor$A\to A_T$. The category $A_T$ has the same objects as $A$, with hom-sets $A_T(a,b) = T(a,b)$. Identities and composition are given by the unit and multiplication of $T$.

Every functor $B \to A$ yields a monad $A(f,f)$ in $Prof$, whose Kleisli object is part of the (bijective on objects, fully-faithful) factorization $B \to A_{A(f,f)} \to A$ of $f$.

Because of this, we can identify a monad on $A$ in $Prof$ with a bijective-on-objects functor $A \to B$.

Remarks

A Kleisli object in a 2-category $K$ is the same as an Eilenberg-Moore object? in $K^{op}$; see opposite 2-category. Kleisli objects for monads in $K^{co}$ can be identified with Kleisli objects for comonads in $K$.

A Kleisli object for a monad $t$ in $K$ can equivalently be defined as a particular sort of weighted2-colimit, namely the lax colimit of the lax functor$\ast \to K$ corresponding to $t$.