Once you know that $R$ has characteristic $2$, it will be a vector space (an algebra, really) over the field $F$ with $2$ elements, of dimension $d$, say. So $R$ has $2^d$ elements.
For each $a \in R$, consider the linear map $T_a : R \to R$ given by $u \to au$. Note that $a \mapsto T_a$ gives a morphism of rings from $R$ to $\operatorname{End}(R)$ (these are the endomorphisms of $R$ as a vector space over $F$). This morphism is injective because $T_a(a) = a$, so its kernel is $0$. If $S$ is its image, then $R$ is isomorphic to $S$.
Since $S$ is commutative, and all of its elements are roots of $x^2-x$, the elements of $S$ can be simultaneously diagonalized, with $0$ and $1$ on the diagonal. There are $2^d$ matrices like this, and $R \cong S$ has $2^d$ elements, so $S$ is the whole ring of diagonal matrices. Choose the (canonical) basis $T_{a_{1}}, \dots, T_{a_{d}}$ such that all $T_{a_{i}}$ have only one $1$ on the diagonal (which corresponds to $T_{a_i}(a_i) = a_i^2=a_i$).
Then with respect to the basis ${a_{1}}, \dots, {a_{d}}$ the ring $R$ has the form you require, because $a_i a_j = 0$ for $i \ne j$.
I guess this is just a version for this case of Stone's representation theorem.