Day 9: Real-World Agentic Systems

This article showcases real-world examples of agentic AI systems – NotebookLM, Perplexity, and DeepResearch – and how they utilize core components like tools, planning, RAG, and memory. While these systems aren't open source, the author breaks down their likely internal workings based on external behavior, demonstrating how these components combine to perform tasks ranging from personal research assistance to complex market analysis. The article also categorizes these systems into agent levels (2-4), highlighting the degree of autonomy and decision-making each exhibits, and encourages readers to experiment with them to better understand agent design.