Proof assistants are wonderful tools to check mathematics but, for most mathematician, their output is exactly one bit of information: true or false. In this talk I present ongoing joint work with Kyle Miller building a tool to extract explanations from formal proofs. In such formally-backed explanations, the reader gets to interactively choose the level of detail.