We believe that the cause why many data science projects fail is that are error-prone and long winding data cleaning processes. We are building the tool - tidyAI - for data scientists that automates data preprocessing and cleaning. tidyAI is built on top of an AI agent, which can adapt to the particular challenges present in the data file. The planner detects the mistakes present in the data that might prevent a future analysis or visualization and generates a list of transformations to apply. The executor takes the top transformation and applies it to the dataset. We do not want to take over the job of a data scientist and analyst, we want to free him up from the grunt work, the work he hates to do, so the data scientists can focus on the work that excites them and that provides value: the analysis.