Research
Peer-reviewed papers and technical white papers from the Minds research team — on synthetic personas, multi-agent creativity, and AI-driven market research.
Research
We publish peer-reviewed papers and technical white papers on the methods behind Minds. Our research covers synthetic persona design, multi-agent systems, creative diversity in LLM outputs, and the future of AI-driven market research.
Latest Papers
- The Spark Effect: On Engineering Creative Diversity in Multi-Agent AI Systems — Persona-conditioned LLM agents deliver a +4.1 point diversity gain over uniform prompts, closing 82% of the gap to human experts.
Research Areas
Synthetic Personas
How do you build AI personas that faithfully represent real-world consumer segments? We study the design, calibration, and validation of synthetic personas for market research.
Multi-Agent Creativity
Single-agent LLM systems converge on homogeneous outputs. Our multi-agent Spark architecture uses persona-conditioned agents to intentionally diversify ideation — producing outputs that approach human-level creative diversity.
AI Evaluation & Benchmarking
LLM-as-a-judge protocols are fast but biased. We develop calibration methods that quantify and mitigate evaluator bias against human gold standards.
Applied Market Research
From concept testing to brand strategy, we study how synthetic panels perform against traditional qualitative research methods — and where the limits are.