A scenario: Imagine Maria driving a wheel-chair to a shopping centre. She needs to get money and then can do her shopping. First she needs to find an ATM she can use. Attached to her wheel chair is the device she uses for accessing such things as ATMs, the washing machine etc. Generally, her chair device, known as the universal remote console (URC), is not online in the sense of being connected to the Internet but it does make wireless connections to local devices. The URC is now within wireless reach of an ATM that might be suitable. Maria uses the discovery metadata from the ATM to determine if it will be suitable for her. This information is provided by the ATM, upon request from her URC, and can be of several forms, minimal or rich, depending upon what the ATM has to offer. In fact, the ATM has only basic information - what it can do in terms of interface features and its location and physical features. This is enough for Maria to determine that she can use her URC to interact with the ATM and get money for her groceries. Maria's URC uses a secure connection to 'talk' to the ATM, telling it that she needs extra time for entering her PIN and similar details, and asking for the cash she wants. As Maria has difficulty remembering how much money she should take out on any particular occasion, she relies on the special 'skin' that the ATM offers to users of smart URCs. This 'skin' helps Maria focus on ATM functionality that helps her choose how much money to take out etc. (She has seen her friend Rose using the same ATM with a different skin that helps her by using her voice input technology.) Maria has to do some grocery shopping but first she wants to treat herself to a new microwave oven. She enters the 'white goods' shop and is able to see how well the various makes of microwave ovens respond to her URC. Some immediately provide her with custom skins that show their features and how to use them; others just have generic skins, and she gets basic information but it is not set out in any specially friendly way, and some make no response to the URC. Maria chooses the one she likes the best in terms of how it communicates with her via the URC and how it will look in her kitchen. Paying for the new oven is so easy, she wonders how she did it in the past - today she just locates the shop's credit payment system, types in the amount, and presses the 'pay' button - done! It's so good to not have to struggle with writing her signature. And it is very satisfying to give the delivery instructions to the shop as part of the payment process, instead of having to write out those boring details. Pleased with her purchase, Maria goes off to the supermarket. So how does this system work? The targets, or products, know about themselves and have one or more ways of presenting this information: * (badly) on a paper label or brochure which suggests they are the best products; * electronically, using generic skins which provide information about all the features normally associated with such a product (depending, of course, on what kind of product it is), or * with special skins that are designed to provide extra information and to be user friendly for different communities of users. It is, of course, metadata that is made available for the shaping of the skins. Currently we are considering the idea of using extended DC metadata to capture essential information - esp. to make a standard that is required of all products sold in the US. Such products include TVs etc but also could include websites .... The extension to DC being considered is a DC-accessibility element that would contain some special accessibility information. We are also considering the user profile for the metadata about the user's needs. And we have a model that has three entities: the target device or service, the user and their URC, and dictionaries that are used for interpretation of the metadata. These dictionaries will probably take the form of metadata registries. The skins can come from the manufacturers or they can come from third parties.