Year: 2026
Type: dissertation
Abstract: This thesis explores whether large language models can help small and medium-sized businesses create circular business model innovations. The research included a literature review and practical tests of prompting workflows using large language models. Based on the theory section, a workflow was created to turn SME information into a structured Business Model Canvas and add circular strategies. For testing, a fictional Finnish SME called NordBox was made up and used. Different prompting methods were tried, including one-shot prompting, source-grounded prompting, step-by-step questioning, and circularization workflows. The study showed that large language models can create clear first drafts of Business Model Canvases and suggest useful circular economy ideas. Source-grounded and structured prompts were usually more reliable than unstructured or very complex prompts. Making prompts more complicated did not always lead to better results. The main differences between methods were in transparency, SME involvement, and how well sources could be traced. Image-based outputs were less reliable than text-based ones because they often missed elements or had formatting problems. This thesis concludes that large language models can help SMEs in the early stages of developing circular business models by organizing information, spotting gaps, and coming up with initial strategy ideas. However, they cannot create fully validated or ready-to-use business models on their own. Human judgment, company context, and checking what is possible are still very important. LLMs should be used as support tools that can give SME users potential avenues of research and can give useful results, especially at the early stages of a circular business model canvas creation process.
Source: Theseus (Ammattikorkeakoulujen)
Author: Niemikotka, Anton
Language: English
Subfield: Strategy and Management
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7162705969