AI-mbidexterity? A qualitative single-case study on how an AI-startup simultaneously pursues exploration and exploitation

Sammanfattning

Organizational ambidexterity is how a firm is able to simultaneously pursue two competing activities: exploitation, the refinement and improvement of existing offerings, and exploration, the search for new opportunities and innovation. While ambidexterity has been extensively studied in larger firms, and increasingly within smaller firms, its empirical application in the context of AI-startups is missing. This is a notable gap, as AI-startups operate in environments characterized by rapid technological change, low entry barriers, and evolving customer demands, which are conditions that intensify the pressure to both deliver short-term results and continuously innovate. This thesis therefore investigates how an AI-startup pursues both exploration and exploitation, and which mechanisms enable this. A qualitative single-case study was conducted on an AI-startup providing B2B AI-assistants, based on 18 semi-structured interviews with employees across commercial, technical, and executive functions. The findings show that the company manages to pursue both activities simultaneously through a contextual ambidexterity mode. Three enabling mechanisms were also identified: (1) leadership, which cultivates an experimentation-tolerant culture, uses quarterly OKRs to adjust focus, and organizes hackathons as dedicated exploratory spaces; (2) AI-tools, which enhance both exploitative efficiency and exploratory capacity, and (3) partnerships, which provide external avenues for scaling existing solutions and co-creating new ones. This study also contributes to ambidexterity literature by illuminating how the support dimension of contextual ambidexterity can be enhanced through AI use, and highlighting an overlooked risk with AI-driven exploration. It also outlines practical implications for managers seeking to establish a better balance between exploration and exploitation within an AI-startup.