Developing in Agentic AI Systems Part 1 of 2
Microsoft Learn
Last updated May 12, 2026
This learning path is the first half of a two-part introduction to designing, deploying, and managing agentic AI systems within the software development lifecycle. Across three modules it covers the foundations of AI coding agents in GitHub — how they plan, act, and improve within a workflow — and how to design agent architectures that separate planning, reasoning, and execution to keep them reliable and controllable. The final module covers tool use and environment interactions, including configuring agent tools, permissions, and MCP (Model Context Protocol) servers, with attention to safety boundaries and traceability.
What you'll learn
- How AI coding agents plan, act, and improve within GitHub workflows
- Integrating agents into the software development lifecycle with clear task boundaries
- Designing architectures that separate planning, reasoning, and execution
- Configuring agent tools, permissions, and MCP servers
- Managing risks, anti-patterns, and traceability of agent-generated work
Frequently asked questions about Developing in Agentic AI Systems Part 1 of 2
Who is Developing in Agentic AI Systems Part 1 of 2 for?
Developers and engineers who want to understand how to design and control agentic AI systems within a software development workflow.
Is Developing in Agentic AI Systems Part 1 of 2 free?
Yes — Developing in Agentic AI Systems Part 1 of 2 is completely free to take.
What are the prerequisites for Developing in Agentic AI Systems Part 1 of 2?
A GitHub account, a basic understanding of AI fundamentals, familiarity with repositories, branches, and pull requests, and general knowledge of CI/CD. No GitHub Copilot subscription required.
Why we suggest this course
For developers and engineers who want to understand how autonomous coding agents fit into a real software process — not just how to prompt one — this path takes a systems view: the agent lifecycle, sensible architecture, and the controls that keep agent-generated work safe and accountable. It treats GitHub as the system of record where that work is tracked and reviewed. One thing to know: the prerequisites are a GitHub account, a basic understanding of AI fundamentals, familiarity with repositories, branches, and pull requests, and general knowledge of continuous integration and continuous delivery (CI/CD) — but no GitHub Copilot subscription is required.