Software agents are very different from mobile robots. For instance, they exist in a world quite unlike the world robots exist in. Also, software agents are not embodied. We investigate how these two differences reflect on agent design principles like situatedness and embodiment. We find that the insights gained from designing mobile robots cannot be applied directly to software agent design. Similarly, in the other direction, insights gained from designing software agents cannot be readily applied to mobile robots. This leads us to a notion of situatedness that is different from the notion of being in the world.