I'm currently studying category theory out of Awodey's excellent online book, but I'm having trouble seeing that the mapping between two product objects is necessarily an iso.
If $P,p_0:P\rightarrow A,p_1:P\rightarrow B$ and $X,x_0:X\rightarrow A,x_1:X\rightarrow B$ are product diagrams for $A$ and $B$ in a category $\mathcal{C}$, then we obtain unique arrows $u_0:X\rightarrow P$ and $u_1:P\rightarrow X$ which yield the identities $$x_0=p_0\circ u_0,$$ $$p_0=x_0\circ u_1,$$ which we can then substitute into each other to obtain $$x_0=x_0\circ u_1\circ u_0,$$ $$p_0=p_0\circ u_0\circ u_1.$$ At this stage the author asserts that these identities, together with the identities $$x_0=x_0\circ1_X,$$ $$p_0=p_0\circ1_P$$ and the uniqueness condition on $u_0$ and $u_1$ yield the desired equalities $u_0\circ u_1=1_P$ and $u_1\circ u_0=1_X$, but I can't quite see it.
If there is some other arrow $u_2:P\rightarrow X$ such that $u_0\circ u_2=1_P$ then I agree we would have a contradiction, since we then have $$p_0=p_0\circ1_P=p_0\circ u_0\circ u_2=x_0\circ u_2$$ and $u_1$ is unique with this property, but is the existence of $u_2$ necessary if $u_0\circ u_1\neq1_P$? Could the identity not hold and there be no such $u_2$?
(There are of course corresponding identities for $p_1$ and $x_1$, but they lead me to the same question.)
Since $P, p_0,p_1$ is a product diagram, any pair of arrows $C\to A, C\to B$ leads through it by a unique $C\to P$.
Now, for the pair $(x_0,x_1)$ this unique arrow is $u_0:X\to P$, and for the pair $(p_0,p_1)$, it is certainly $1_P$. But, taking the compositions, also $u_0\circ u_1:P\to P$ leads $(p_0,p_1)$ through itself.
This means, by uniqueness, $u_0\circ u_1=1_P$.