Implementing and Designing AI-Powered Chatbots: Organisational Perspectives on Data Quality, Key Drivers, and Barriers in Medium and Large Swedish Enterprises

Sammanfattning

This bachelor’s thesis investigates how medium and large Swedish enterprises perceive data quality during the design and implementation of AI-powered chatbots for internal use. Through five semi-structured interviews with professionals directly involved in the chatbot development, the study explores organisational perspectives on data accuracy, completeness, and reliability, as well as key drivers and barriers influencing the adoption. Findings indicate that trust in the chatbot’s output is heavily dependent on data quality, integration with existing systems, and transparency. Organisational support, user involvement, and clear use case definitions emerged as critical drivers, while lack of technical infrastructure, data governance, and user understanding posed significant barriers. The study applies the TOE and PACT frameworks to analyse how technological, organisational, and user-centric factors interact during implementation. By highlighting the role of end-user trust and practical alignment with organisational needs, this research contributes to a deeper understanding of chatbot integration in enterprise environments. The findings offer practical recommendations for organisations seeking to implement internal chatbots more effectively.