Book by Eric Ries. Official website.

Describes a way to deal with uncertainty when creating a new product. One important aspect reitered over the book is that it applies of course to startups but recipes can also be used in established companies to continue creating disruptive products.

The main mantra is to follow a suite of build-measure-learn cycles: starting with a clear hypothesis, build the minimum viable product (which may be much much more limited than what one actually thinks is required) to test it against real customers and then check whether the hypothesis was valid. Those cycles should be kept as short as possible to get feedback from actual customers, and overall metrics should show if the current direction makes sense to realize the mission of the startup or if building another product would make more sense.