Agentic Hyperautomation : Implementation in Power Platform

Abstract

This thesis examined the concept of agentic hyperautomation and demonstrated its capabilities in a practical implementation within the Microsoft Power Platform. Its objective was to redesign the hours reporting process of the Human Resources department of a company. The process was highly manual, error-prone, and time-consuming for human actors. The aim was to evaluate the agentic approach in action and how it could reduce manual effort, improve quality, simplify approvals, and increase process efficiency but always keeping human actors in the loop. The study consists of a theory section where the concepts were explained in detail. Then the implementation was carried out using the components available in the Microsoft Power Platform, such as Power Automate, Dataverse, Copilot Studio. An agent was designed to interact with users, retrieve and validate information, and, additionally, trigger automation. A robot was responsible for the execution of the repetitive tasks, previously done by humans manually. The results showed that most of the tasks of the process were successfully automated and could run mostly without human intervention. Human actors were able to interact with the agent during the process through natural language to handle and supervise its flow. Substantially, the redesign improved efficiency and reduced workload. In conclusion, the thesis showed that agentic hyperautomation enhance effectively the business processes then with the help of low-code technologies made easy its implementation and conversational artificial intelligence extended its scope. The collaboration between agent, robot and humans made the business process flexible and adaptable to the human needs. The results suggested the agentic hyperautomation has strong potential adoption in companies that want modern business processes and operations.