Course Overview : Professional Agentic AI Architect Course Overview

This advanced course prepares learners to architect and implement enterprise-grade agentic AI solutions using Google Cloud capabilities. 

Learners will progress from configuring low-code agentic workflows and coding agents to building custom agents using the Agent Development Kit, integrating enterprise knowledge through RAG and vector retrieval, implementing agent memory and identity, and orchestrating multi-agent workflows using protocols such as MCP and A2A. 

The course also covers the complete production lifecycle of agentic applications, including evaluation, deployment, scaling, observability, troubleshooting, security, governance, guardrails, and human-in-the-loop controls. 

The curriculum is aligned with the Professional Agentic Architect certification exam domains and the tools identified as being in scope in the official exam guide.  

After completing Agentic AI Architecture course, participants will be able to:

  • Design and configure agentic workflows using low-code tools.
  • Create effective system instructions and prompt templates.
  • Connect agents securely to enterprise data sources.
  • Configure coding agents with MCP servers, Skills, Extensions, Hooks, and tools.
  • Use coding agents for code generation, refactoring, optimization, and vulnerability remediation.
  • Build custom agents using Agent Development Kit (ADK).
  • Configure sessions and agent memory.
  • Design RAG and vector retrieval solutions.
  • Configure agent identity and permissions.
  • Integrate agents with APIs, databases, SaaS applications, and external tools.
  • Orchestrate multi-agent workflows using MCP and A2A.
  • Implement sequential, parallel, and graph-based agent workflows.
  • Build evaluation datasets and continuous agent evaluation pipelines.
  • Deploy and scale production agentic workloads.
  • Monitor and troubleshoot agent performance.
  • Implement authentication, secure tool execution, and access boundaries.
  • Apply governance, guardrails, Model Armor, Agent Gateway, and human-in-the-loop controls.

Upcoming Batches

Loading Dates...

Key Features: Professional Agentic AI Architecture

  • Enterprise Agent Architecture: Design production-ready autonomous AI solutions.  

  • Custom Agent Development: Build agents using ADK and modern agent development patterns.

  • Coding Agents: Configure and customize coding agents using MCP, Skills, Extensions, Hooks, and subagents.  

  • RAG & Enterprise Knowledge: Connect agents to enterprise data using retrieval and vector search.  

  • Multi-Agent Architecture: Implement MCP and A2A-based agent collaboration.  

  • Agent Evaluation: Create golden datasets, evaluation sets, and continuous evaluation workflows.  

  • Production Deployment: Deploy and scale agents using Agent Runtime, Cloud Run, and GKE.  

  • Security & Governance: Implement Agent Identity, Agent Gateway, Model Armor, OAuth 2.0, policies, and guardrails.

  • Hands-On Learning: Practical labs and architecture scenarios throughout the course.  

  • Capstone Project: Design and implement a production-ready multi-agent enterprise solution.  

Who should Attend?

  • AI/ML Engineers
  • Generative AI Engineers
  • AI Architects
  • Cloud Architects
  • Solution Architects
  • Software Architects
  • Application Developers
  • ML Engineers
  • DevOps Engineers
  • Platform Engineers
  • Technical Leads
  • AI Solution Designers
  • Professionals preparing for the Professional Agentic Architect certification

Prerequisites for the Agentic AI Architecture Learning course:

  • Strong programming fundamentals.  
  • Experience with Python or another programming language.  
  • Familiarity with APIs and JSON.  
  • Basic understanding of LLMs and generative AI.  
  • Familiarity with cloud concepts.  
  • Experience with Git and software development workflows.  
  • Basic understanding of databases and data retrieval is recommended.  
  • Prior exposure to AI agents is beneficial but not mandatory.  

Why choose CloudThat as your training partner?

  • Google Cloud & AI Expertise: Learn from trainers experienced in cloud and AI technologies.
  • Industry-Experienced Trainers: Practical architecture and development scenarios are incorporated into the training.  
  • Hands-On Learning: Learners build agents rather than only studying theoretical concepts.  
  • Architecture-Focused Approach: Emphasis on selecting appropriate models, runtimes, data architectures, orchestration patterns, and security controls.  
  • Enterprise Use Cases: Training incorporates real-world enterprise agent scenarios.
  • Certification Alignment: Curriculum follows the Professional Agentic Architect exam domains.  
  • Interactive Learning: Includes demonstrations, labs, architecture discussions, quizzes, and troubleshooting exercises.  
  • End-to-End Learning: Covers the complete agent lifecycle—from design and development through evaluation, deployment, monitoring, security, and governance.  

Learning Objectives of the Agentic AI Architect Course

  • Agent Architecture: Design scalable and reliable agentic AI architectures.  
  • Coding Agents: Configure and customize coding agents using MCP, Skills, Extensions, Hooks, and subagents.  
  • Custom Agent Development: Build custom agents using ADK. 
  • Enterprise Knowledge: Implement RAG, vector retrieval, and enterprise data integration.  
  • Agent Orchestration: Build multi-agent systems using MCP and A2A.  
  • Agent Memory: Configure sessions, state, and memory for persistent agent experiences.  
  • Evaluation Excellence: Develop golden datasets and continuous agent evaluation workflows.  
  • Production Readiness: Deploy, scale, monitor, troubleshoot, and optimize agentic workloads.  
  • Security & Governance: Implement Agent Identity, Agent Gateway, Model Armor, access boundaries, and guardrails.  
  • Responsible AI: Design secure agent behavior with appropriate safety mechanisms and human oversight.

Course Outline: Professional Agentic AI Architect Download Course Outline

Lecture Content

  • Introduction to agentic AI
  • Agentic workflows and autonomous systems
  • Agent architecture fundamentals
  • Agent states and transitions
  • State-based workflows
  • Pages and transition routes
  • Event handlers
  • Configuring agents using low-code tools
  • System instructions
  • Prompt templates
  • Few-shot prompting
  • Agent behavior configuration
  • Enterprise agent use cases

Learning Objectives

  • Explain core concepts of agentic AI.
  • Design state-based agent workflows
  • Configure agent behavior using system instructions.
  • Create prompt templates for agent workflows.
  • Connect agents to enterprise information.

Activities

  • Create a basic agentic workflow. 
  • Configure states and transitions.
  • Create system instructions.
  • Build a low-code enterprise agent.
  • Quiz.

Lecture Content

  • Introduction to coding agents
  • Coding agents vs traditional AI assistants
  • Configuring coding agents
  • MCP servers
  • Agent Skills
  • Tools and tool permissions
  • Extensions and plugins
  • Hooks and rules
  • Subagents
  • Secure coding environments
  • Coding agents in sandboxed environments
  • Code generation
  • Code refactoring
  • Runtime optimization
  • Application vulnerability remediation
  • Coding-agent best practices

Learning Objectives

  • Configure coding agents for software development.
  • Extend coding agents using MCP and Skills.
  • Use coding agents to analyze and modify code.
  • Implement secure coding-agent workflows.
  • Automate development and remediation activities.

Hands-on Lab: AI Coding Assistant

  • Developer
  • Coding Agent
  • Repository Analysis
  • Code Generation / Refactoring
  • Testing
  • Security Validation

Lecture Content

  • Enterprise coding-agent architecture
  • Creating Skills
  • Plugins
  • Extensions
  • Hooks
  • Rules
  • Subagents
  • Custom tools
  • Agents CLI
  • Agents vs human mode
  • Standardizing developer workflows
  • Enterprise coding workflows
  • Scaling coding agents 
  • Governance of coding agents
  • Optimizing deployed agents

Learning Objectives

  • Customize coding agents for enterprise development.
  • Build reusable Skills and Extensions.
  • Configure Hooks and rules.
  • Create specialized subagents.
  • Govern and optimize coding-agent workflows.

Hands-on Lab: Build an Enterprise Coding Agent

  • Repository analysis
  • Code review
  • Unit-test generation 
  • Security checks
  • Documentation generation

Lecture Content

  • Agent Development Kit overview
  • Agent architecture
  • Creating custom agents
  • Agent instructions
  • Tools
  • Tool calling
  • Agent execution
  • Sessions
  • Context
  • Agent memory
  • Managed sessions
  • Memory Bank
  • State management
  • Agent lifecycle
  • Model selection
  • LLM vs SLM
  • Open-source vs proprietary models
  • SaaS vs self-hosted models
  • Model selection based on cost, security, and performance

Learning Objectives

  • Build custom agents using ADK.
  • Select appropriate models for agent workloads
  • Configure sessions and memory.
  • Manage agent state and context
  • Design agents according to cost, security, and performance requirements.

Hands-on Lab: Build a Custom ADK Agent

  • User
  • ADK Agent
  • Model
  • Response

Lecture Content

  • Enterprise knowledge integration
  • Retrieval-Augmented Generation
  • RAG architecture
  • Document ingestion
  • Chunking
  • Embeddings
  • Vector retrieval
  • Similarity scoring
  • Reranking
  • Vector databases
  • Agent Retrieval
  • Vector Search
  • RAG Engine
  • Multimodal enterprise data
  • Data access permissions
  • Grounded agent responses
  • RAG troubleshooting

Learning Objectives

  • Design enterprise RAG architectures.
  • Configure vector-based retrieval.
  • Select appropriate retrieval strategies.
  • Improve agent grounding and response quality.
  • Integrate enterprise knowledge securely

Hands-on Lab: Build a RAG-Powered Enterprise Agent

  • Enterprise Documents
  • Processing
  • Embeddings
  • Vector Retrieval
  • Reranking
  • Agent
  • Grounded Response

Lecture Content

  • Agent tools and capabilities 
  • Agent Identity
  • Agent permissions
  • Identity propagation
  • Agent Registry
  • Google Cloud MCP Servers
  • Custom integration layers
  • Database integrations
  • API integrations
  • SaaS integrations
  • Remote MCP servers
  • Secure agent-to-tool communication
  • OAuth 2.0
  • Tool authorization
  • Least-privilege access

Learning Objectives

  • Configure secure agent identities.
  • Control agent permissions.
  • Integrate agents with enterprise applications.
  • Connect agents to APIs and databases.
  • Secure agent-to-tool communication.

Hands-on Lab: Secure Agent-to-Tool Integration

  • Agent
  • Agent Identity
  • Authorization
  • MCP / API
  • Enterprise System

Lecture Content

  • Multi-agent architecture
  • Agent specialization
  • Agent handoffs
  • Agent2Agent protocol
  • MCP-based agent interactions 
  • Sequential agents
  • Parallel agents
  • Graph workflows
  • Agent coordination
  • Agent policies
  • Agent Registry
  • Agent Runtime
  • Agent Identity
  • Multi-agent failure handling
  • Agent communication patterns

Learning Objectives

  • Design multi-agent architectures.
  • Implement agent handoffs.
  • Select sequential, parallel, or graph-based workflows.
  • Coordinate specialized agents.
  • Apply policies to multi-agent systems.

Hands-on Lab: Multi-Agent Research Assistant

  • User
  • Orchestrator
  • Final Agent
  • Answer

Lecture Content

  • Why agent evaluation is different
  • Agent evaluation lifecycle
  • Test sets
  • Golden datasets
  • Prompts and expected responses
  • Edge cases
  • Tool execution evaluation
  • Retrieval evaluation
  • Response quality
  • Custom autoraters
  • ADK evaluation
  • Evaluation services
  • Continuous evaluation
  • Regression testing
  • Agent quality metrics
  • Evaluation troubleshooting 

Learning Objectives

  • Create agent evaluation datasets.
  • Evaluate agent responses and tool execution.
  • Measure retrieval quality.
  • Build continuous evaluation pipelines.
  • Identify agent quality regressions.

Hands-on Lab: Evaluate an Agent

  • Golden Dataset
  • Agent
  • Evaluation
  • Response Tool Retrieval
  • Evaluation Report

Lecture Content

  • Production agent architecture
  • Agent Runtime
  • Cloud Run
  • GKE
  • Deployment strategies
  • Scaling agents
  • Runtime selection
  • Performance optimization
  • Latency optimization
  • Tool invocation latency
  • Agent reasoning loops
  • Agent drift
  • System failures
  • Cloud Logging
  • Cloud Trace
  • Agent observability
  • Cost optimization
  • Production troubleshooting

Learning Objectives

  • Select an appropriate agent runtime.
  • Deploy and scale production agents.
  • Monitor agent performance.
  • Troubleshoot agent failures.
  • Optimize latency, reliability, and cost.

Hands-on Lab: Deploy and Monitor an Agent

  • Custom Agent
  • Agent Runtime / Cloud Run / GKE
  • Monitoring
  • Logs Trace Metrics
  • Optimization

Lecture Content

  • Agent security architecture
  • Authentication
  • OAuth 2.0
  • Secure tool execution
  • Agent-to-tool authentication
  • Agent Identity
  • Principal Access Boundary
  • Agent Gateway
  • Agent Registry
  • Model Armor
  • Data protection
  • Identity propagation 
  • Agent policies
  • Safety frameworks
  • Guardrails
  • Human-in-the-loop
  • Prompt injection
  • Data exfiltration
  • Sensitive information protection
  • Secure agent behavior
  • Governance and compliance

Learning Objectives

  • Design secure agentic architectures.
  • Implement authentication and authorization.
  • Control agent permissions.
  • Protect enterprise data.
  • Implement guardrails and safety controls.
  • Configure governance policies.
  • Incorporate human approval into high-risk workflows.

Hands-on Lab: Secure Enterprise Agent

  • Agent
  • Agent Gateway
  • Authentication
  • Guardrails
  • Agent Identity
  • Model Armor
  • Enterprise Tools

Learners design and implement an enterprise-grade multi-agent solution incorporating the major capabilities covered throughout the course.

Solution Requirements

  • Custom ADK agents
  • Gemini model
  • Agent sessions and memory
  • RAG / enterprise knowledge
  • Vector retrieval
  • Agent Identity
  • MCP integration
  • A2A-based agent communication
  • Multi-agent orchestration
  • Agent evaluation
  • Production deployment
  • Observability
  • Security and governance
  • Human-in-the-loop controls

Reference Architecture

  • User
  • Agent Gateway
  • Orchestrator
  • MCP/A2A
  • Agent Runtime
  •     Evaluation + Monitoring
  •     Security + Governance

Certification Details: Professional Agentic AI Architect

    • Upon successful completion of the training, participants will be prepared with the knowledge and practical skills aligned to the Professional Agentic Architect certification exam.
    • CloudThat Course Completion Certificate will be awarded to learners who successfully complete the training.
    • The official exam guide describes the Professional Agentic Architect as a practitioner who designs and manages autonomous AI-driven workflows while considering reliability, performance, cost, security, and scalability.

Click to Zoom

Select Course date

Loading Dates...
Add to Wishlist

Course ID: 32504

Course Price at

Loading price info...
Enroll Now

FAQs for Agentic AI Architect Training

This is an advanced course focused on designing, developing, evaluating, deploying, securing, and governing enterprise agentic AI solutions.

No. The focus is specifically on agent architecture and agentic application development, including ADK, MCP, A2A, RAG, agent memory, evaluation, orchestration, security, and governance.

Yes. Learners build custom agents and multi-agent workflows using ADK and related agent technologies.

Yes. MCP is covered both for coding-agent integrations and custom agent/tool integrations.

Yes. Sequential, parallel, graph-based workflows, agent handoffs, MCP, and A2A are covered.

Yes. Enterprise RAG, embeddings, vector retrieval, similarity scoring, reranking, Agent Retrieval, Vector Search, and RAG Engine are covered.

Yes. Agent Runtime, Cloud Run, GKE, scaling, monitoring, troubleshooting, and optimization are included.

Yes. Agent Identity, OAuth 2.0, Agent Gateway, Model Armor, access boundaries, guardrails, and human-in-the-loop controls are covered.

Yes. Learners create test sets, golden datasets, evaluation workflows, and continuous evaluation pipelines.

Yes. Learners design an enterprise multi-agent architecture incorporating development, RAG, orchestration, evaluation, deployment, observability, and security.

Talk to Us