In pervasive Internet access, the quality of presentation often needs to be adjusted according to the instantaneous network bandwidth and the hardware capability of the user's device. Current web transformation mechanisms, being unable to adapt dynamically to the available network bandwidth and data requirement, result in uniform loss of quality, and incur additional processing overhead for every user. Hence, it is important to estimate the network conditions of each individual user dynamically, in order to perform appropriate transformation on the web content for different users to suit their constraints. However, the WWW is a dynamic and diverse environment with great heterogeneity among its users, servers, networks, and data. This makes it difficult for us to discover the diverse network constraints of different users. In this paper, we propose a bandwidth sensitive content transformation proxy solution to deliver the best-fit presentation to web surfers. Transformation to web content will be done according to the amount of latency saved, the availability of the current network bandwidth, and the workload of the servers due to transformation. To illustrate its potentials, automatic web data compression is used as an example on how adaptive content transformation will be performed and optimized.