Agentic AI
Systems where Large Language Models (LLMs) autonomously navigate multi-step reasoning, tool usage, and self-directed workflows to solve complex problems without constant human intervention.
Core Patterns
5
Implementations
11
Avg. Reliability
—
Coming SoonEvaluator-Optimizer Agent
An iterative refinement loop where an 'Evaluator' provides granular feedback on an 'Optimizer’s' output until quality thresholds are met.
ReAct Agent
Interleaves chain-of-thought Reasoning with Action execution, enabling LLMs to dynamically plan, act, and observe in a loop.
Tool Use Agent
Augment an LLM with callable external tools — APIs, code interpreters, databases — so it can take actions and retrieve real-time information beyond its training data.
Plan-and-Execute
Separate high-level planning from step-by-step execution: one LLM call produces a structured plan, then individual executor calls carry out each step, with replanning triggered by unexpected results.
Multi-Agent Orchestration
Coordinate a network of specialised AI agents under an orchestrator, where each agent owns a distinct capability or domain and agents communicate through structured messages.