COMPEL Glossary / agentic-ai
Agentic AI
Agentic AI refers to artificial intelligence systems capable of taking autonomous actions in the world, making decisions, using external tools, and pursuing multi-step goals with minimal or no human intervention at each step.
What this means in practice
Unlike traditional AI that responds to individual queries, agentic AI can plan sequences of actions, interact with other systems via API calls, and adapt its approach based on intermediate results. For organizations, agentic AI introduces fundamentally new governance challenges because these systems can take consequential actions independently, potentially at speed and scale that outpaces human oversight. In the COMPEL framework, agentic AI governance is addressed through dedicated articles in Modules 2.4, 2.5, 3.3, 3.4, and 4.3, covering topics from the autonomy spectrum and delegation frameworks to multi-agent orchestration and agentic failure taxonomies.
Why it matters
Agentic AI introduces fundamentally new governance challenges because these systems can take consequential actions independently, at speed and scale that outpaces human oversight. Organizations deploying agentic AI without appropriate governance frameworks risk financial loss, regulatory violations, and reputational damage from autonomous systems acting outside intended boundaries. The governance gap between traditional AI and agentic AI is one of the most urgent challenges in enterprise AI today.
How COMPEL uses it
Agentic AI governance is addressed across multiple COMPEL stages through dedicated content spanning the autonomy spectrum, delegation frameworks, multi-agent orchestration, and agentic failure taxonomies. The Model stage designs governance architectures specific to agentic AI risk. The Produce stage implements controls including action space constraints and kill switches. The Evaluate stage monitors agent behavior against delegated authority boundaries, and the Learn stage captures agentic AI incident patterns.
Related articles in the Body of Knowledge
- Evaluating Agentic AI: Goal Achievement and Behavioral Assessment
- Agentic AI Architecture Patterns and the Autonomy Spectrum
- Agentic AI Maturity Assessment: Extending the 18-Domain Model
- Operational Resilience for Agentic AI: Failure Modes and Recovery
- Designing Measurement Frameworks for Agentic AI Systems
- Agentic AI Cost Modeling: Token Economics, Compute Budgets, and ROI
- Enterprise Agentic AI Platform Strategy and Multi-Agent Orchestration
- Agentic AI Governance Architecture: Delegation, Authority, and Accountability
Related Terms
Other glossary terms mentioned in this entry's definition and context.