In the business automation world, APIs are everywhere. They provide access to enterprise tools such as customer relationship management solutions, or custom automations such as unattended RPA bots that automate repetitive tasks. Unfortunately, they may not be accessible to the business users who need them but are not equipped with the necessary technical skills to leverage them. Most recently, chatbots are becoming the go-to medium to make automation software accessible to business users. Since API specifications aren’t written with a chatbot use in mind, additional work is needed to make APIs accessible through a natural language interface. Making this process scalable to many APIs requires an automated data training pipeline for intent recognition models, a crucial component within chatbots to understand natural language utterances from users. More accurate intent recognition models lead to better user experience and satisfaction. Prior work proposed approaches to extracting intents from OpenAPI specifications. However, the resulting models tend to be brittle due to weaknesses in training data. In this work, we propose a data augmentation approach based on paraphrasing using large language models and propose a system to generate sentences to train intent recognition models. Experimental results highlight the effectiveness of our approach. Our system is deployed in a real world setting.