Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn to build agentic AI systems in this comprehensive crash course that defines agents as software entities using LLMs to perceive environments, make decisions, and execute actions to achieve specific goals. Explore the critical distinction between static workflows and dynamic agentic systems, understanding how LLMs serve as a reasoning "brain" to decompose tasks at runtime. Master essential components including system prompts, tools, and memory through practical Python demonstrations while comparing architectural patterns such as Supervisor and Swarm. Discover the three pillars of AI (algorithms, data, and compute), examine the spectrum of autonomy, and understand when to use agents versus traditional workflows. Build hands-on experience progressing from single LLM calls to custom Python agents, adding memory and history, and implementing agents with frameworks like LangChain. Analyze multi-agent systems, evaluate agentic performance using code-based, LLM-as-a-Judge, and human evaluation methods, and address current challenges including hallucinations, cost, and debugging. Examine real-world incidents through the AI Incident Database, assess career impacts across different job categories, and explore the evolution toward Software 3.0 development paradigms. Investigate emerging interoperability protocols like MCP (Model Context Protocol), A2A, and AGUI while looking ahead to world models and the future of Artificial Machine Intelligence (AMI) beyond current LLM capabilities.
