How can we use the power of machine learning on our own data, in our documents, our CMS? This talk explores possibilities to harness this power and shows that it can easily be done with your own data using embeddings, Redis and a bit of PHP to build a search - or chat - feature to find content by meaning rather than mere keywords