Systems that support mobile agents are increasingly being used on the global Internet. An important application that is considered for these agents is electronic commerce, where agents roam the World Wide Web in search of goods for their owners. In these applications, an agent moves along some itinerary in order to search for the best offer for the good sought by the user. The problem with this approach is that malicious providers on the agent's itinerary can damage the agent, tamper with the agent so that the agent itself becomes malicious, or forward the agent to any arbitrary provider that might not be on the agent's itinerary. In this presentation we will primarily address the question how an agent can safely follow some pre-defined itinerary. We will identify the problem of trust as a major issue in this context and describe a trusted and tamper-proof hardware that can be used to enforce a policy. Based on this policy, we will show how the agent can take advantage of it in order to achieve the desired goal.