The rapid growth of digital text emphasises the necessity to create and build efficient summary tools to find and extract information in short form. Text overviews are typically handled using extraction techniques in natural language processing, which include choosing a portion of the original text to capture the core concept of the subject. Unlike the abstractive technique, it is utilised less than the extractive method, but outcomes are considerably closer to the human summary. In our study, we concentrate on the automated extractive kind of summary. In our study, we utilised Word2Vec, a collection of lexical integration models from Word Embedding. Word2Vec turns the word into a vector, and these segments are the derivatives of the word in our instance a 100 segment vector. We also utilised the CNN and auto-encoder. We also used the CNN. We use our model on Amazon Food Reviews datasets and assess rebuilt paragraphs using standard metrics (such as ROUGE). The aim of this research was to automatically generate the accurate and best summary.