Web is the boundless source of information and no one is able to process the vast amount of new documents published on the web every day, even with filtering out the documents the user is not interested in. However, most of the recent web documents are blog posts, news and other documents with the author information established. Each author who is also the receiver of web documents possesses their own personal agent that delivers trust information related to other authors as well as rank data for each new document. Trusts and ranks available for agents are exchanged between them and in this way new authors and new web documents can be easily assessed. Based on the general concept of Web of Trust the new idea of Personal Web of Trust and its application to local ranking method for web documents is proposed in the paper.