Agentic AI Planning and Reasoning Design Patterns

Course details

There are foundational design techniques commonly used to build agentic AI solutions with basic planning and reasoning capabilities. This course goes beyond those techniques to explore a range of intermediate and advanced design approaches focused on specialized agent architectures that can be designed to carry out diverse forms of runtime planning and reasoning logic, as well as self-correction logic and adaptive learning capabilities that do not rely on model training. Join LinkedIn Top Voice and best-selling author Thomas Erl as he covers a set of design patterns that reveal sophisticated and powerful ways in which agentic AI solutions can be tailored to perform different types of planning and reasoning logic based on different business requirements and performance constraints.

Concepts

Introduction

Introduction
What you need to know
Scope of design pattern coverage

Planning and Execution Design Patterns

Planning and execution design patterns - Overview
Plan-and-execute
Concurrent execution optimizer
Reasoning without observation
Planner-critic-refiner

Search and Self-Correction Design Patterns

Search and self-correction design patterns - Overview
Language agent tree search
Second-pass verification
Dynamic self-discover

Memory, Skill, and Adaptive Action Design Patterns

Memory, skill, and adaptive action design patterns - Overview
Episodic and procedural memory
In-context learning
Adaptive tool orchestration

Conclusion

Next steps