This module applies the platform knowledge from Module 1 to a complete workspace blueprint design exercise, then extends that work into the communication skills required to get the blueprint approved. You apply a structured three-step design methodology to produce a production-ready Foundry workspace blueprint for a realistic enterprise scenario, then practice presenting and defending that design to a skeptical security stakeholder in a simulated executive review.

This module establishes the agent design landscape, introducing the three core agent types as essential context, before focusing exclusively on task agent design. You will develop the ability to identify when a task agent is the right choice and produce a production-ready operational specification with triggers, inputs, outputs, and escalation logic for a real enterprise scenario.

This module develops design skills for the two remaining agent types—autonomous agents and prompt/response agents. You will configure the five foundational pillars of autonomous agent design, specify the key design components of prompt/response agents, and apply both sets of skills through a refinement exercise that integrates all three agent types.

This module develops the architectural design skills needed to structure a multi-agent system—defining the orchestrator-specialist topology, assigning agent roles and responsibilities, and specifying the A2A inter-agent communication parameters that govern handoffs. Learners apply these skills to a complex enterprise process that requires coordination among multiple specialist agents.

This module extends the multi-agent topology design from Module 1 into the operational layer—designing the Foundry Workflow logic that orchestrates agent execution, with error handling, retry logic, and observability instrumentation. Learners apply these skills first through a pre-production architecture review coach dialogue, then through a structured workflow failure diagnosis—identifying production gaps, specifying remediations, and communicating findings to an engineering team as a practicing architect would before any enterprise deployment goes live.

This module elevates learners' use of generative AI tools, moving beyond AI-assisted drafting to AI-assisted architectural validation. Learners apply two advanced techniques: using a GenAI tool (like Microsoft Copilot) as an adversarial critic to stress-test their own multi-agent topology before deployment, and using a GenAI tool to run a pre-mortem temporal inversion; forcing the AI to construct a future failure narrative that surfaces the blind spots architects are most likely to ignore when they are proud of their own designs. The result is a more rigorous, production-ready architecture and a validation capability that directly differentiates senior architects from implementation specialists.

Learners step into the role of Lead AI Architect for a mid-size commercial insurance company deploying its first enterprise multi-agent AI system. Rather than producing isolated deliverables for each short course, learners produce a single integrated Architecture and Sign-off Package, covering Foundry workspace design, agent operational specifications, multi-agent governance and A2A security design, and a workflow stabilization assessment. The result mirrors the complete pre-deployment architecture package a practicing AI solutions architect produces before any enterprise multi-agent system goes live.