As I believe, one of the most significant developments in mathematics has been the study of dualities. There are many standard ones, like Hilbert's Nullstellensatz, Schemes, Gelfand duality, or Tannaka duality (on the dualities note, you may see Lecture 1 of Scholze's Gestalten notes). We will see the Tannaka (Grothendieck–Saavedra–Deligne–Milne) duality for a commutative Hopf algebra, however, in an incomplete way here. Other similar posts include theQuill20 and theQuill21.
An affine group scheme is defined over a field $k$ to be the representable functor $G: \mathsf{CAlg_k} \to \mathsf{Grp}$. The coordinate ring on $G$, denoted by $\mathcal{O}(G)$, forms a commutative Hopf algebra structure (this can be realized using Yoneda's Lemma). The converse also exists, that given a commutative Hopf algebra $A$, we can find the corresponding affine group scheme, which is the spectra $Spec A(R) = Hom_k(A, R)$. The representation of an affine group scheme is given by a natural transformation of functors $\rho: G \to Aut_V$ where $Aut_V(R) = Aut_R(V \otimes_k R)$ is another functor valued in $\mathsf{Grp}$. If rank $n$ vector space $V$ is free, then $\rho: G \to GL_n$ as $Aut_V \cong GL_n$. Moreover, the comodules over Hopf Algebra, in this case, we will take $\mathcal{O}(G)$, are defined as usual.
The claim is that there is a canonical equivalence between the category of representations of the affine group scheme $G$ and the category of comodules over the Hopf algebra $\mathcal{O}(G)$
$$ \mathsf{Rep_k(G)} \simeq \mathsf{Comod_k(\mathcal{O}(G))}.$$ Recall, if there exists a symmetric (tensor), faithful, exact, k-linear fiber functor $\omega: \mathsf{T} \to \mathsf{Vect_k}$, then we call $\mathsf{T}$ a (neutral) Tannakian category $\mathsf{T} \cong \mathsf{Rep_k(G)}$ which is a rigid, k-linear, tensor category and there exists a fundamental group associated to the Tannakian category which is the affine group $G= Aut^{\otimes}(\omega)$. (One can also realize that the tensor structure on $Rep_k(G)$ is induced by the Hopf algebra.) Overall, the duality between affine group schemes and commutative Hopf algebras is $$ \{\text{affine group schemes over k} \} \leftrightarrow \{ \text{commutative Hopf Algebra over k}\}$$ $$ G \to \mathcal{O}(G)$$ $$ Spec A \leftarrow A$$
Given a fiber functor $\omega: \mathsf{T} \to \mathsf{Vect_k}$, we can get a Hopf algebra $H$ from co-end construction (we will discuss (co)ends shortly on this blog) $$H = \int^{X \in \mathsf{T}} \omega(X)^\vee \otimes_k \omega(X).$$ The matrix coefficients from the coend $H$ can be realized to be associated with the representative functions on $G$ for $\omega$ a forgetful functor. Given a monoidal functor $\omega: \mathsf{T} \to \mathsf{Vect_k}$, one says that $\mathsf{T}$ is equivalent to the category $\mathsf{Comod_k(\mathcal{O}(G))}$ and under this Tannakian hypothesis and ($T \cong \mathsf{Rep_k(G)}$), we have $H \cong \mathcal{O}(G)$. More appropriately, the Hopf algebra structure on $H$ is given by the coend construction, where the (co)multiplication and (co)unit are given by the tensor structure of $\omega$. (I apologize for having rushed this part for this post; please see this, this, and this text for more appropriate definitions and follow-ups.)