Co-Creation of Knowledge in Marketing: Human-AI Collaboration in the Construction of Proto-Personas

How can we develop an understanding of consumers in a specialised market where structured data are scarce and much of the available knowledge resides in practitioners’ experience?

This study explores that question through human–AI collaboration in the construction of proto-personas: provisional representations of ideal customers that help teams organise insights, formulate hypotheses and guide further investigation.

The research brings together marketing education, knowledge management and the Brazilian bonsai market. At ESAG/Udesc, students interacted with John Jobs Naka, a generative AI agent configured with specialist knowledge drawn from interviews, technical publications and secondary data. They used these conversations to interpret consumer profiles and develop proto-personas through group work.

A partnership grounded in human expertise

The study examines this collaboration through Knowledge Management 5.0 and the SECI model: Socialisation, Externalisation, Combination and Internalisation. Expert experience was recorded and incorporated into the agent’s knowledge base. Students then questioned, selected, interpreted and synthesised the information, connecting it with their own learning.

The agent served as a mediator within this process. Human interpretation, critical judgement and responsibilityremained central to the construction of marketing knowledge.

What the students reported

The evaluation obtained 67 valid responses from a population of 71 students, a response rate of 94.4%.

98.5% recognised the agent as an effective partner or mediator in knowledge construction. The clarity of its responses received a score of 4.19/5, while recommendability reached 8.57/10.

The qualitative analysis identified three themes: cognitive mediation, empathy and humanisation, and the integration of tacit and explicit knowledge.

These findings describe students’ perceptions in this educational context. They do not constitute independent validation of the agent’s factual accuracy or proof that the proto-personas represent the wider bonsai customer population. The paper identifies the absence of external expert validation, the restricted knowledge corpus and limited opportunities for refinement as limitations.

Why this research matters

The study offers an exploratory approach to making specialist knowledge available for learning and marketing investigation. It suggests that generative AI, when grounded in a carefully prepared knowledge base, may support iterative enquiry and the development of consumer hypotheses in markets with limited structured data.

For educators, the approach creates opportunities for applied learning. For marketing teams, it offers a starting point for further customer research and validation.