Why this is in the programme
As agentic systems become more complicated, the problem moves from individual model calls towards orchestration: deciding which component does what, how state moves through a workflow and where control should remain deterministic.
Course
Agentic AI with LangGraph, CrewAI, AutoGen and BeeAI — IBM
The course examines agentic frameworks and workflow patterns using LangGraph, CrewAI, AutoGen and BeeAI.
Role in the experiment
The purpose is not to commit the eventual system to any of these frameworks.
Instead, this stage will provide enough practical experience to evaluate whether orchestration or multiple agents genuinely improve the eventual solution, and to recognise situations in which they merely add complexity.