This question concerns the theorem that says for any bounded sesquilinear form on a hilbert space $H$, $u:H\times H \to \mathbb{F}$ we can uniquely associate it with bounded operators $A,B$ such that $u(h,k) = \langle Ah,k\rangle = \langle h,Bk\rangle$ for all $h,k\in H$.
The proofs I've seen for this proceed the same way in defining a functional $L_h: H \to \mathbb{K}$ by $L_h(k) = \overline{u(h,k)}$, applying the Riesz representation theorem to get $f\in H$ such that $\langle k,f\rangle = L_h(k) = \overline{u(h,k)}$, then defining $A$ to be the operator satisfying $Ah=f$ whence $\langle Ah,k \rangle = \overline{\langle k,Ah\rangle} = \overline{\langle k,f \rangle} = u(h,k)$.
What I'm confused about is how this uniquely associates a single $A$ to $u$ rather than associating a different $A$ to every pair of $h$ and $k$, since the Riesz representation theorem would give a different $f$ for different $u$ and for different $h$ different $A$ would make $Ah=f$.