Summarizer is a tool for creating short text summaries. It utilises text extraction method, i.e. the output consists of sentences from the original text. The tool uses a number of machine learning algorithms, including neural networks, linear regression, Bayesian networks and decision trees. The output sentences are chosen based on different signals, such as the length of the sentence, its position in the text structure and properties of the words it contains. The system was trained specifically for newspaper articles in Polish. It is possible, however, to adjust it for other kinds of documents and languages.