Despite decades of intervention, smallholder agriculture often shows limited impact when technologies overlook farmers’ real needs. A major challenge is that farmer feedback, while abundant, is difficult to analyse systematically in a timely and consistent way, so critical perspectives are often lost before informing design. To address this gap, this study applies AI-assisted thematic analysis to real raw qualitative data extracted from farmer-led workshops in Cambodia. Going beyond simple structuring, we summarise 39 bullet-style farmer quotes into four concise, interpretable categories, completing the process in less than a minute. Using Natural Language Processing, including sentence embeddings and unsupervised clustering algorithms, we identify latent patterns in the text and then apply Generative AI via the OpenAI API in Python to generate meaningful theme titles that reflect the lived realities of smallholders. The resulting themes reveal critical physical, technical, and socio-economic challenges farming communities face, with emphasis on barriers disproportionately affecting women and youth. This hybrid approach transforms raw qualitative input into summarised, structured, visual insights and offers a proof-of-concept method for turning community-generated feedback into actionable knowledge. The findings inform early-stage design of electric two-wheel tractors (e2WTs) by integrating diverse insights and centring on farmer-focused innovation for sustainable mechanisation in Cambodia.