For a few short months at the end of 2001 I was a member of a young and very ambitious start-up called the User Advocacy Group (UAG) who were, at the time, working with Amey to help them improve the Manchester and Croydon Tram systems.
UAG were a usability company - but they didn't undertake usability testing of websites. They didn't even do HCI, as such. Instead, they looked at the usability of any complex system where people and technology routinely interacted - like a modern tramway - with the same suite of analytic probes and synthetic tools.
The methodology used was highly unusual and bespoke, combining state-of-the-art systems thinking with modern ethnographic and market research techniques. The company was formed of a bright, eclectic mix of people - and the work was facinating. For me, coming from primarily a web background, the whole notion of contextual design of non-web systems was a revelation - and has provided a wonderful foil to grandiose or far-fetched claims asssociated with web usability.