Sumários
Agentic AI Patterns
12 Dezembro 2025, 13:00 • Luis Manuel Ferreira Fernandes Moniz
Agentic AI Patterns define structured ways to build systems where Large Language Models act as autonomous agents. Instead of relying on single prompts, these patterns introduce roles, workflows, and control mechanisms that improve reliability, safety, and scalability. Examples include Planner and Task Decomposition for multi-step goals, ReAct and Reflexion for reasoning and self-correction, Router and Orchestrator for coordinating multiple agents, Parallel and Multi-Agent CoT for robustness, LTS + RAG for memory and grounding, and Guardrails + Evaluator for safety and quality. Together, these patterns transform LLMs into predictable, collaborative, and trustworthy agentic systems suitable for real-world tasks.
Agentic AI
5 Dezembro 2025, 13:00 • Luis Manuel Ferreira Fernandes Moniz
Agentic AI refers to AI systems in which Large Language Models go beyond text generation to act as goal-driven agents. An agentic system can plan, make decisions, use tools, interact with other agents, and adapt its behavior based on feedback and memory. Unlike traditional LLM usage, Agentic AI emphasizes autonomy, control, and interaction with an environment. Key elements include planning and execution loops, reasoning and self-evaluation, memory (short- and long-term), communication between agents, and safety mechanisms. Agentic AI enables complex, multi-step tasks and collaborative problem solving, forming the foundation for scalable, reliable, and trustworthy autonomous AI systems.