We present a general set-up for inflation in string theory where the inflaton field corresponds to Wilson lines in compact space in the presence of magnetic fluxes. T-dualities and limits on the value of the magnetic fluxes relate this system to the standard D-brane inflation scenarios, such as brane-antibrane inflation, D3/D7 brane inflation and different configurations of branes at angles. This can then be seen as a generalised approach to inflation from open string modes. Inflation ends when the Wilson lines achieve a critical value and an open string mode becomes tachyonic. Then hybrid-like inflation, including its cosmic string remnants, is realized in string theory beyond the brane annihilation picture. Our formalism can be incorporated within flux-induced moduli stabilisation mechanisms in type IIB strings. Also, contrary to the standard D-brane separation, Wilson lines can be considered in heterotic string models. We provide explicit examples to illustrate similarities and differences of our mechanism to D-brane inflation. In particular we present an example in which the eta-problem present in brane inflation models is absent in our case. We have examples with both blue and red tilted spectral index and remnant cosmic string tension $G\mu \lesssim 10^{-7}$.

We consider a cosmological scenario within the KKLT framework for moduli stabilization in string theory. The universal open string tachyon of decaying non-BPS D-brane configurations is proposed to drive eternal topological inflation. Flux-induced `warping' can provide the small slow-roll parameters needed for successful inflation. Constraints on the parameter space leading to sufficient number of e-folds, exit from inflation, density perturbations and stabilization of the Kahler modulus are investigated. The conditions are difficult to satisfy in Klebanov-Strassler throats but can be satisfied in T^3 fibrations and other generic Calabi-Yau manifolds. This requires large volume and magnetic fluxes on the D-brane. The end of inflation may or may not lead to cosmic strings depending on the original non-BPS configuration. A careful investigation of initial conditions leading to a phenomenologically viable model for inflation is carried out. The initial conditions are chosen on the basis of Sen's open string completeness conjecture. We find time symmetrical bounce solutions without initial singularities for k=1 FRW models which are correlated with an inflationary period. Singular big-bang/big-crunch solutions also exist but do not lead to inflation. There is an intriguing correlation between having an inflationary universe in 4 dimensions and 6 compact dimensions or a big-crunch singularity and decompactification.