AI Agent Taxonomy
Every AI agent
has a weak point.
We mapped every major architecture - perception inputs, decision mechanisms, output actions, and the governance failure modes unique to each. Anchor8 governs all of them.
17
agent architectures
100% governed
- Tap any type to expand
17 types
Perception
Text, documents, RAG/vector search, API responses, conversation history (200k+ token context)
Decision Making
Chain-of-thought, ReAct, Plan-and-Execute, task decomposition, tool selection
Actions
API calls, code execution, email/CRM/ticket creation, report generation
Governance Challenges
HallucinationPrompt injection"Agentwashing"Sensitive data leakage
Your agent type is covered.
Anchor8 monitors decisions, enforces policies, and generates immutable audit trails across all 17 agent architectures - from edge neuromorphic chips to LLM multi-agent pipelines.