An exact AdS/CFT duality

It has recently been proposed that type IIB string theory on AdS3 x S3 x T4 with one unit of NS-NS background flux is exactly dual to the 2d CFT that is described by the symmetric orbifold of T4. Starting from general considerations about the AdS/CFT duality, I shall explain why one may expect such a duality to hold. I shall then also sketch the various arguments in favour of this claim: the full (single-particle) spectrum matches between the two descriptions (including all non-BPS sates), and the structure of the dual CFT correlation functions emerges naturally from the string world-sheet. (This is based on joint work with Lorenz Eberhardt and Rajesh Gopakumar.)