In the realm of business automation, digital assistants/chatbots are emerging as the primary method for making automation software accessible to users in various business sectors. Access to automation primarily occurs through APIs and RPAs. To effectively convert APIs and RPAs into chatbots on a larger scale, it is crucial to establish an automated process for generating data and training models that can recognize user intentions, identify questions for conversational slot filling, and provide recommendations for subsequent actions. In this paper, we present a technique for enhancing and generating natural language conversational artifacts from API specifications using large language models (LLMs). The goal is to utilize LLMs in the "build" phase to assist humans in creating skills for digital assistants. As a result, the system doesn't need to rely on LLMs during conversations with business users, leading to efficient deployment. Experimental results highlight the effectiveness of our proposed approach. Our system is deployed in the IBM Watson Orchestrate product for general availability.

The fields of business process analysis and process mining (PM) analyze business operations to identify, validate, improve, and automate business processes. Most information about business processes is available as unstructured or semi-structured data in IT systems implementing them. Examples of the information include design documents, XML-format configurations, system logs, and database schema descriptions. The information focuses on the IT algorithms for implementing the processes and does not directly map to the business description of the processes. Advances in Natural Language Processing (NLP) techniques related to semantic tagging, topic modelling, and text classification present opportunities to analyze unstructured data. NLP tasks such as text classification rely on an annotated corpus suitable for the domain. The availability of corpus annotated with implementation-specific tags is a well-known limitation. This paper addresses the challenge of mapping annotations from language and domain-level (generic) descriptions of processes into implementation-specific process data. We present a transfer learning-based approach trained on a corpus of annotated domain-level text and semantic tags. We demonstrate that such a learning technique can effectively annotate implementation-level process information. We further compare the use of state of the art Skip-gram, GloVe, ELMO, and BERT-based learning models in implementing our framework.