Agentic AI

Definition

Agentic AI is aI systems that pursue goals through planning, tool use, memory, and feedback loops rather than single-shot inference. In defense applications, it turns models into bounded workers that can decompose tasks, call tools, and coordinate with other agents. The hard part is goal drift, hidden tool misuse, prompt injection, and poor escalation behavior, especially when systems are deployed across contested links, coalition boundaries, and mixed human-machine teams. KhanBMS treats it as an orchestrator of trusted KhanBMS modules, not a direct authority over effects, tying the concept back to modular command, edge execution, and auditable authority.

Reference attributes

Layer

agent architecture

Operational value

Turns models into bounded workers that can decompose tasks, call tools, and coordinate with other agents

Primary risk

Goal drift, hidden tool misuse, prompt injection, and poor escalation behavior

KhanBMS role

An orchestrator of trusted KhanBMS modules, not a direct authority over effects

Related terms

• LLM Orchestration Layer
  Middleware that routes models, prompts, tools, memory, retrieval, policy, and telemetry across AI workflows.
• Tool-Use Agents
  Agents that call external APIs, databases, simulators, sensors, or effectors to accomplish tasks.
• Policy Guardrails
  Deterministic and model-assisted controls that constrain what AI systems may say, decide, or execute.
• Mission Planning Agent
  AI agent that helps generate, revise, and monitor mission plans under commander intent and constraints.