Natural Language Interaction Protocol (NLIP)

The advent of large language models (LLMs) has made an interactive natural language interaction feasible between machines in a manner that did not exist before. An implication is that a natural language interface can replace many mobile applications that are used today. Just like the advent of the browser in 1990s simplified technology by replacing a plethora of client-side applications with a single standard application, a common natural language interaction protocol can potentially replace the plethora of mobile applications that exists today, providing a universal application layer protocol. Convergence to a universal application layer protocol would bring significant benefits to all segments of society - consumers can use a single application for various interactions, businesses will have a simpler maintenance burden for their IT infrastructure, and integration among different businesses can be streamlined. A set of researchers from several different universities and industries have come together to define this protocol, and address the research challenges involved in defining and using it. This protocol must support authentication, secure communication, multi-modal interaction, exchange privacy and trustwor- thiness policies, among others. This protocol must be open-source and developed in a community environment, along with freely available implementations. The initial specification of the protocol is being built using a REST interface over HTTPS using JSON messages for interactions. In this paper, we will provide the motivation, requirements, use-cases, an initial specification and the current state of NLIP.