A method and system manages content delivery to a multi-modal client device. The system may include a session manager logically connected between a server and the client device. The session manager can maintain a record of the state of a user's session, and if the user or the client device switches from an initial presentation mode to a second presentation mode, the session manager can use the record to effect a seamless transfer from the initial presentation mode to the subsequent presentation mode. The session manager may also transcode content delivered to the client device so that the content is compatible with the presentation mode selected by the user.