The AI Anthropology Toolkit is an open-source set of computational tools for qualitative research, designed to implement the principles of Multi-Agent Ethnography. The toolkit provides three specialized AI agents: a Qualitative Codebook Builder that generates structured codebooks from research data, an Interview Transcript Semantic Chunker that segments transcripts into analytically meaningful units, and a Coding and Thematic Analysis Agent that applies codes and develops themes across a corpus. Each agent is configured through epistemological prompting, which allows researchers to set explicit analytical commitments before analysis begins.
The toolkit has gone through multiple versions: standalone Jupyter notebooks (V1), MCP-orchestrated agents integrated with Claude (V2), and a web application (V3). It is released on GitHub and archived on Zenodo, making it available for anthropologists and qualitative researchers to use and adapt. The project serves as an empirical demonstration of AI Anthropology in practice, showing how anthropological knowledge and analytical craft can be formalized into computational infrastructure while preserving human interpretive authority over research findings.
Matt Artz developed the toolkit as part of his PhD research at Aalborg University’s Techno-Anthropology Lab. It is one of the core deliverables of his work in AI Anthropology and demonstrates concepts including machine knowing, AKG construction, and the AI Anthropology Lifecycle.