John Savill's Technical Training explains how to choose between Agent Builder, Copilot Studio, and Microsoft (Azure AI) Foundry for building AI agents, using decision criteria like code complexity, model selection, deployment targets, lifecycle management, observability, evaluations, safety, tools/knowledge, multi-agent needs, memory, audience, and cost control.

Picking the right Agent Builder solution

John Savill's Technical Training walks through a practical decision process for choosing an agent-building approach across Agent Builder, Copilot Studio, and Microsoft Foundry / Azure AI Foundry, depending on how much code you want to write, what you need to operate, and where you plan to deploy.

Video chapters

• 00:00 - Introduction
• 00:17 - Quick way to decide
• 02:02 - Organizational maturity progression
• 05:19 - Building agents
• 05:39 - Agent Builder, Copilot Studio and Microsoft Foundry
• 06:39 - Code complexity
• 10:23 - Model selection
• 13:01 - Deployment target
• 14:27 - Lifecycle management
• 16:09 - Observability
• 17:43 - Evaluations
• 21:13 - Safety
• 22:39 - Tools and knowledge
• 24:04 - Multi-agent
• 25:29 - Memory
• 27:11 - Audience
• 27:42 - Cost control
• 29:18 - Summary, box meals, restaurants and really good kitchens
• 31:09 - Close

What this video helps you decide

Quick decision criteria

The video proposes a fast way to decide based on key constraints, then expands into deeper trade-offs:

• Organizational maturity progression: how teams typically evolve from simpler tooling to more configurable/operationally capable platforms.
• Code complexity: how much custom code and control you need.
• Model selection: how flexible you need to be in choosing models.
• Deployment target: where the agent needs to run.
• Lifecycle management: how you plan to version, update, and manage the agent over time.
• Observability: how you’ll monitor behavior, failures, and performance.
• Evaluations: how you’ll test and measure quality.
• Safety: guardrails and risk management.
• Tools and knowledge: what integrations and knowledge sources the agent needs.
• Multi-agent: whether you need multiple cooperating agents.
• Memory: how the agent should persist context.
• Audience: who the agent is for and how that affects requirements.
• Cost control: cost management considerations as part of choosing the platform.

Summary framing

The video closes with a metaphor-style summary (“box meals, restaurants and really good kitchens”) to explain when a simpler vs more capable builder is appropriate.

Reference materials linked by the author

• Whiteboard diagram: https://github.com/johnthebrit/RandomStuff/raw/master/Whiteboards/WhichAgentBuilder.png
• Table version: https://github.com/johnthebrit/RandomStuff/raw/master/Images/WhichAgentTable.png

Related learning links (from the description)

• Recommended Learning Path for Azure: https://learn.onboardtoazure.com
• Certification Content Repository: https://github.com/johnthebrit/CertificationMaterials
• Weekly Azure Update playlist: https://youtube.com/playlist?list=PLlVtbbG169nEv7jSfOVmQGRp9wAoAM0Ks
• Azure Master Class playlist: https://youtube.com/playlist?list=PLlVtbbG169nGccbp8VSpAozu3w9xSQJoY
• DevOps Master Class playlist: https://youtube.com/playlist?list=PLlVtbbG169nFr8RzQ4GIxUEznpNR53ERq
• PowerShell Master Class playlist: https://youtube.com/playlist?list=PLlVtbbG169nFq_hR7FcMYg32xsSAObuq8
• Certification Cram Videos playlist: https://youtube.com/playlist?list=PLlVtbbG169nHz2qfLvPsAz9CnnXofhmcA
• Mentoring Content playlist: https://youtube.com/playlist?list=PLlVtbbG169nGHxNkSWB0PjzZHwZ0BkXZZ
• FAQ: https://savilltech.com/faq
• Channel subscribe link: https://www.youtube.com/channel/UCpIn7ox7j7bH_OFj7tYouOQ?sub_confirmation=1

Notes and limits

Only the video description (chapters and links) was provided; the detailed spoken content isn’t included in the input. The structure above preserves what’s explicitly available without inventing additional guidance beyond the listed chapter topics.