In order to coordinate their activities, embedded systems must implement means to share data with one another. While robust network connectivity among them is necessary to achieve this goal, in practice it is by no means sufficient. By analogy, advanced Internet applications are not implemented from IP datagrams alone, but rather employ a variety of richer services (e.g., DNS, databases, ...) built over this basic primitive. We believe that implementing richer services in networks of embedded systems is a research area that demands significantly more attention in order to enable the development of embedded systems for a variety of mission-critical applications. In fixed-infrastructure and well-provisioned networks like the Internet, richer services are typically supported with trusted servers or server clusters. However, embedded networks might preclude such an implementation for at least three reasons. First, there may not be a component in the network with the resources to serve access requests for many clients. Second, the network latency and congestion that results from all clients accessing a single location may be prohibitive. Third, due to the challenging environments in which embedded networks may operate, there may not be a node that is sufficiently trustworthy to be relied upon for authoritatively providing services to other nodes in the network. While analogs of these problems exist in fixed-infrastructure networks, embedded systems exacerbate several difficult issues, including the need to continually re-evaluate service-provisioning decisions if nodes move, and to adapt as nodes depart the system due to failure or disconnection. In this talk I will summarize techniques we have developed for implementing robust distributed services in environments characteristic of embedded networks. Our techniques offer ways to disperse service load across many network elements while mitigating the extra latency and/or network congestion that this can induce, and without placing trust in individual nodes to authoritatively serve others. We will additionally discuss new research challenges in this space.