
Create your first Copilot chat agent in Microsoft Copilot Studio with BananaAssist, set its PNG icon under 72 kilobyte, and explore topics, triggers, and responses.
Explore how variables and logic drive a Copilot Studio chat, using an orchestration engine, labeled variables, and conditional prompts to tailor vacation policy responses for employees and contractors.
Learn to use entities to turn messy user responses into clean data, such as age or country, and store it in variables within a playground topic for testing.
Explore Copilot Studio system topics, focusing on conversation start, quick replies, and media like images to customize a friendly chat bot for three daily use topics.
Discover how conversational boosting routes unknown intents to generative answers, leveraging all knowledge sources, handling variables, and using escalate and fallbacks to test robust chat behavior.
Recap the shift from a question-answer parrot to a reader that pulls answers from documents. Learn to connect SharePoint as a knowledge source and control generative answers and conversational boosting.
Enable AI orchestration and let Copilot decide which topic handles the conversation. Watch generative AI use user intent and knowledge sources, producing variable, sometimes confident or muddled answers.
Compare ai orchestration with classic orchestration, decide early between predictability and flexible knowledge-heavy conversations, and apply Banana Assist 2.0 in a corporate policy context.
Organize your AI instructions with clear purpose, general guidelines, skills, and step-by-step procedures to improve orchestration. Apply error handling, interaction examples, and follow-up to build reliable chat agents and testing.
Learn to test your CoPilot chat agent inside CoPilot Studio using the evaluation tool, with single response tests, quick and full question sets, and knowledge sources versus topics.
Explore how Copilot Studio builds AI agents from topics, triggers, variables, entities, and knowledge sources to craft intelligent conversations and test performance.
AI is everywhere, but building a useful AI agent is very different from watching a polished marketing demo.
This course is designed for citizen developers, business users, and anyone who wants to learn Microsoft Copilot Studio by building real-world AI agents from scratch. No coding experience is required.
You'll start by understanding what Copilot Studio really is, what it can do well, and where its limitations begin. From there, you'll build a complete AI-powered HR assistant step by step while learning the core building blocks of Copilot Studio, including Topics, Triggers, Variables, Entities, Knowledge Sources, and Generative Answers.
Once the foundations are in place, you'll explore AI Orchestration, learn when to use it, when to avoid it, and how to maintain control when AI starts making decisions on your behalf. You'll also learn how to create effective instructions, organize your agent's behavior, and use knowledge sources to generate accurate and relevant answers.
Most importantly, you'll learn how to properly test and evaluate your AI agent. Building an agent is easy. Building one that consistently works is where the real challenge begins.
By the end of this course, you'll have the confidence to design, build, test, and deploy intelligent AI agents using Microsoft Copilot Studio and understand the reasoning behind every design decision along the way.