Meet your instructor
Hugo Herrera BEng, MCP, MCTS is a Lead Dynamics 365 and AI Solutions Architect and the author of Microsoft Power Platform Solutions Architect’s Handbook.
With 15 years of experience designing and shipping enterprise solutions on the Microsoft stack, Hugo now builds production agentic systems using Copilot Studio, Azure AI Foundry, and the Microsoft Agent Framework. His current work includes a multi-agent platform that he uses every day.
Hugo has also sat the new AB-100 Agentic AI exam, so this workshop is grounded in both current Microsoft certification expectations and practical enterprise implementation.
What makes this workshop unique:
AI agents are moving quickly from experimental pilots to real enterprise workloads. Organizations are no longer just asking whether an agent can be built. They are asking where agents fit, which tools to use, how to ground them in trusted business knowledge, how to evaluate their performance, and how to put the right controls in place before they are used in production.
This five-hour live build-along workshop gives you a practical, end-to-end view of what it takes to design and deliver AI agents on the Microsoft stack. Through guided labs, you will explore how Copilot Studio, Azure AI Foundry, and the Microsoft Agent Framework each support different parts of the agent development journey, from low-code business agents to evaluated specialist agents and code-first multi-agent workflows.
Instead of focusing on isolated demos, the workshop follows the full delivery path: selecting the right tool, grounding agents on relevant knowledge, connecting them to tools and workflows, evaluating their outputs, introducing orchestration patterns, and applying production-readiness checks. The emphasis is on understanding how agentic systems behave in realistic enterprise conditions, including the points where quality, safety, governance, and human oversight become critical.
You will compare the strengths and trade-offs of each Microsoft agent tool and learn how they fit together in a broader enterprise AI strategy. Across the session, the focus is on practical decision-making as much as implementation, helping you understand not only how to build agents, but how to assess whether they are reliable, controlled, and ready for real-world use.
Reserve your place and build enterprise AI agents across Copilot Studio, Azure AI Foundry, and the Microsoft Agent Framework in one intensive live session!
Who should attend the workshop:
This workshop is designed for professionals who need to understand how AI agents fit into real Microsoft-stack delivery, not just at demo level but as part of enterprise solution design. It is especially relevant for solution architects, technical consultants, senior developers, technical leads, Power Platform and Dynamics 365 professionals, Azure and Microsoft 365 practitioners, and anyone involved in evaluating, designing, or delivering agentic AI solutions for business use cases.
You do not need to be a specialist AI engineer to benefit from the workshop. The Copilot Studio and Azure AI Foundry modules are portal based and suitable for non-developers, while the Agent Framework module introduces the code-first patterns in a guided, accessible way. All prompts, commands, and snippets are provided in advance for copy and paste, with completed checkpoints, companion guidance, replay support, and the session recording available so both coders and non-coders can follow the learning path confidently.
Pre-requisites:
To take part in the live labs you will need:
- An Azure subscription able to deploy the workshop model (a free account upgraded to pay-as-you-go works).
- A Copilot Studio trial, signed up with a work or school account (a personal email is not accepted).
- A Microsoft 365 tenant (a one-month Business Basic trial works).
- Visual Studio Code with either .NET or Python installed for the code-first module.
Licences must be valid on the day. The joining pack includes trial activation guidance, a five-minute environment check that proves a live model completion, and a licence pulse check three days before the workshop, so an expired trial is caught early rather than on the day.
On a trial you can build each agent and validate it in the test panel during the session. Publishing an agent to a channel needs a full licence, so Hugo runs the publish step live and you take home a guide to repeat it once you are fully licensed.
What you will learn:
- Choose the right Microsoft agent tool for different enterprise requirements
- Build a grounded Copilot Studio agent with instructions, knowledge, a custom tool, and test validation
- Create and evaluate an Azure AI Foundry agent grounded on documents, then fix issues based on evaluator results
- Implement a code-first agent workflow using the Microsoft Agent Framework, custom tools, memory, and C# or Python
- See and follow a two-agent handoff with a human approval gate on a sensitive tool such as payments, mirroring Microsoft’s documented patterns (coders implement it live; non-coders follow with the recording and solved code)
- Apply production readiness checks across safety, observability, identity, cost control, and AB-100 alignment
What you will walk away with:
- Three working Microsoft AI agents built across Copilot Studio, Azure AI Foundry, and the Microsoft Agent Framework
- A reusable tool-choice framework to decide which Microsoft agent tool to use, when, and why
- A production readiness checklist for reviewing safety, observability, identity, cost, and deployment gaps
- An AB-100 alignment map connecting the workshop patterns to the certification skills areas
- The full lab repository with completed solutions at every checkpoint
- A copy-paste coursework pack with every prompt, command, snippet, and replay guide needed to rebuild the labs later