Written by the AgileFever editorial team. Program details were checked against the live program pages in October 2026. We run these courses. Skills, labs and exam coverage come from the three program pages, checked on October 28, 2026.
If you have to pick one cloud, the skills are the same and the names are different. This post lines them up so you can see what you would build, what changes between platforms, and what to learn first.
The six skills on each cloud
| Skill | Azure (Microsoft Foundry) | AWS (Amazon Bedrock) | Google Cloud (Agent Platform) |
|---|---|---|---|
| 1. Set up for AI work | Foundry project, Entra ID and managed identity, quotas, budget alert | Model access, IAM least-privilege role, quotas, budget alert | Project and billing, IAM and service accounts, ADK install, budget alert |
| 2. Deploy models | ML model on Azure Machine Learning and a foundation model, both on managed endpoints | ML model on a SageMaker AI endpoint, foundation models through the Bedrock Converse API | ML model on Agent Platform endpoints, Gemini and Model Garden models |
| 3. RAG and guardrails | Azure AI Search, content safety, prompt-injection shields, evaluators | Bedrock Knowledge Bases, Guardrails, Bedrock evaluations | RAG Engine and Vector Search, Model Armor, evaluation |
| 4. Build agents | Foundry portal and the Microsoft Agent Framework in Python, MCP tools | Bedrock Agents with Lambda tools, and Strands Agents in Python with MCP | Agent Development Kit (ADK) in Python, Memory Bank, MCP tools |
| 5. Multi-agent systems | Orchestration patterns, human approval, A2A, bring in LangGraph or CrewAI | Supervisor and specialists in Strands, human approval, A2A, bring in LangGraph or CrewAI | Coordinator and specialists in ADK, human approval, A2A, Agent Registry |
| 6. Operate | Deploy to Foundry Agent Service, tracing with Application Insights, evaluation, cost controls | Deploy to AgentCore Runtime, observability with CloudWatch, guardrails, IAM policies | Deploy to Agent Runtime, tracing and evalsets, Model Armor, cost controls |
Identical on all three pages
Additional discounts are available. Contact us to know more.
Where the clouds differ
- AWS is the only course that teaches two agent routes: the managed Bedrock Agents and a code-first route with the Strands Agents SDK.
- Google Cloud is code-first throughout, built on ADK, and includes a short low-code comparison.
- Azure builds the same agent twice, once in the Foundry portal and once in Python.
Dates, price and prerequisites
| Item | Azure | AWS | Google Cloud |
|---|---|---|---|
| Next cohort listed | November 16 to 19, 2026 | November 23 to 26, 2026 | November 30 to December 3, 2026 |
| Schedule | Monday to Thursday, 9:30 AM to 12:30 PM Eastern | Monday to Thursday, 9:30 AM to 12:30 PM Eastern | Monday to Thursday, 9:30 AM to 12:30 PM Eastern |
| Fee (US) | $199 | $199 | $199 |
| Monthly EMI | From about $9 | From about $9 | From about $9 |
| Vendor exam it builds toward | Microsoft AI-103 | AWS Certified Generative AI Developer, Professional (AIP-C01) | Google Cloud Professional Agentic Architect (in beta as of October 2026) |
All three ask for the GenAI and Agentic AI BootCamp or equivalent experience building agents, working Python including API and SDK calls, and a cloud account where you may create resources.
What the exam preparation covers, and what it does not
Each page is explicit that the course covers part of the exam. The Azure course covers the generative AI and agent domains of AI-103 but not computer vision, text and speech analysis or information extraction. The AWS course covers Bedrock, RAG, Guardrails, agents and deployment at a working level, while the Professional exam goes deeper on architecture, security, governance and cost. The Google Cloud course covers custom agents (about a third of the exam) and deployment, but not low-code tools or coding agents for application development. Each page points to an exam-prep add-on.
Which to choose
Start with the cloud your employer runs. If it runs more than one, pick the one where your next project will be deployed. The three pages are built the same way, with the same six steps, so what you learn on one gives you a map of the others. For a side-by-side of the platforms, read Azure vs AWS vs Google Cloud for AI agents, and for the Azure course in detail, the Azure curriculum breakdown. The architecture view is in the AI Architecture Cost Model and Checklist. The course pages are Enterprise AI Agents on Azure (Microsoft Foundry), Enterprise AI Agents on AWS and Enterprise AI Agents on Google Cloud.
Questions people ask
Do I need to take all three?
No. The six skills are the same on each cloud, so one course teaches the pattern and you apply it to another platform with its documentation.
Is any of these a vendor certification?
No. You receive an AgileFever course completion certificate. Each course is aligned to a vendor exam and covers part of it.
Are sessions recorded?
Yes. Live sessions are recorded.
