We provide a detailed discussion of the main theoretical and phenomenological challenges of quintessence model building in any numerically controlled regime of the moduli space of string theory. We argue that a working quintessence model requires a leading order non-supersymmetric (near) Minkowski vacuum with an axionic flat direction. This axion, when lifted by subdominant non-perturbative effects, could drive hilltop quintessence only for highly tuned initial conditions and a very low inflationary scale. We then present a type IIB 4D string model with stabilised moduli which is able to describe the history of the universe from inflation to quintessence. The underlying Calabi-Yau volume is controlled by two moduli which are stabilised by perturbative effects. The lighter of them drives Fibre Inflation at a large energy scale. The two associated axions are ultra-light since they are lifted only at the non-perturbative level. The lighter of them can drive quintessence if its decay constant is large enough to prevent quantum diffusion during inflation from ruining the initial conditions. The right dark energy scale can be obtained via a large suppression from poly-instanton effects. The heavier axion gives a negligible contribution to dark matter since it starts oscillating after matter-radiation equality. If instead none of the two axions has a large decay constant, a mild alignment allows the lighter axion to drive quintessence, while the heavier can be at most a few percent of dark matter due to isocurvature and UV bounds. In both cases dark matter can also come from either primordial black holes or the QCD axion.