We develop and introduce NLSI (Natural Language Standing Instructions), an English-language dataset in which every example consists of a conversation
between the user and a dialogue agent, accompanied by a collection of standing instructions (user profile) and a sequence of API calls reflecting user
intents.