Focus groups are useful for acquiring topic-relevant thoughts, behaviors, attitudes, and emotions. The most useful focus groups are typically large enough to ensure diverse views, small enough to ensure contributions from each participant, and carefully moderated to allow participants to challenge, clarify, and otherwise engage with one another’s ideas. With the ideal focus group requiring considerable time, preparation, and funding, artificially intelligent (AI) agents may be useful as proxies for human participants when resources to obtain qualitative insights are scarce. In this paper, we provide an overview of recent work in this field, as well as development and feasibility testing of Predictive Evaluation of Responses to Change, Engagement, and Psychological Trends (PERCEPT), a novel research and development effort to conduct AI focus groups with relevant cognitive, demographic, and psychographic human persona characteristics. Our approach incorporates best practices in multi-agent large language modeling (LLMs), Tree-of-Thought (ToT), retrieval-augmented-generation (RAG), sentiment analysis, persona development, prompt engineering, and simulated moderator guidance, among other methods. For the purposes of feasibility testing, a small-n research design was implemented to evaluate an artificial focus group’s representativeness of a real-world target population, as well as value and completeness of artificially generated insights. Challenges to adopting this technology, including ethical replication of human biases, are also discussed.