The company's aim is to help customers optimize anything, whether it's an ad campaign or a formula for shaving cream. In fact, co-founder and CEO Scott Clark (pictured above) told me that his newer customers include Pembient, a bioengineering company that creates synthetic animal materials - it's using SigOpt to develop a synthetic version of the rhino's horn.

'The technology's really for anyone testing things on a trial-and-error basis,' Clark said. 'That could be someone in a lab running experiments that take dozens to hundreds of hours, or it could could be in a machine learning setup that requires dozens or hundreds of computer hours.'

In the case of a physical experiment, you'd manually enter both the experiment and the results, then SigOpt could tell you what variation to test next. Clark said it can automatically help you identify the 'points of highest improvement' while balancing between the need for 'exploration' (i.e., trying out totally new ideas) and 'exploitation' (drilling down on the areas that are already working).

Clark also argued that SigOpt doesn't compete A/B testing tools - in fact, it offers one-click integration with Optimizely.

The technology is based on the Metric Optimization Engine, a tool that Clark developed while working on Yelp's ad targeting team. The MOE is open source, but SigOpt is building premium products and services around it. Clark said this is 'democratizing' optimization making it accessible to a wider range of teams.

The seed funding comes from Andreessen Horowitz and Data Collective. Among other things, it's allowing SigOpt to hire its first engineers and move into its first San Francisco office.