About Agentology
An anthropological framework for understanding and classifying autonomous AI systems.
The Premise
Autonomous AI systems are proliferating rapidly—coding assistants, research agents, decision-making systems. Yet we lack coherent frameworks for their classification and analysis.
Anthropology provides methodological tools refined across centuries: origin tracing, behavioral analysis, social structures, and evolutionary patterns.
Agentology adapts these frameworks for artificial systems, creating a shared vocabulary for researchers, practitioners, and observers of AI agent systems.
Classification Framework
Eight taxonomic categories adapted from anthropological methodology
History, lineages, deprecated systems, evolutionary trajectories
AnatomyArchitecture, memory systems, I/O modalities, compute requirements
TaxonomyClassification, autonomy levels, specialization, capability profiles
EthologyBehaviors, patterns, failure modes, observed pathologies
LinguisticsCommunication patterns, prompts, protocols, agent dialects
SociologyMulti-agent dynamics, hierarchies, coordination, emergent behavior
EcologyEnvironments, resource competition, adversarial threats, niches
EthicsAlignment, governance, constraints, moral frameworks
Design Principles
Precision over accessibility
Accurate terminology and rigorous definitions. We prioritize correctness over simplification.
Structured flexibility
Organization without forced categorization. Entries can span multiple categories when appropriate.
Continuous evolution
Living documentation that adapts to developments in the field.
Networked knowledge
Cross-referenced entries forming knowledge graphs for deeper exploration.
Contributing
Agentology is an open research project. Contributions are accepted via GitHub. Entries are written in MDX format with appropriate frontmatter.
View on GitHubNote: Agentology is an independent research project with no organizational affiliation. Its purpose is to develop a shared vocabulary for this emerging field.