Model Architectures · Practitioner
Agent Architectures & Tool-Calling Loops
Build reliable agents: the ReAct loop, state and memory, termination conditions, error recovery, and when a workflow beats an agent.
- 1 hours lesson
- advanced level
- Engineering Track
- Narrated presentation
What you’ll learn
- ReAct: reason, act, observe loops
- Workflows vs. autonomous agents
- Iteration/token budgets and stop conditions
- Error feedback and loop detection
- Scratchpad memory and state persistence
- Compounding error rates and human checkpoints
Inside this lesson
A narrated, slide-by-slide presentation with 1 prediction exercise, 1 knowledge check, 3 videos. You answer as you go, and the platform checks your work.