Exam objective 5.1 asks you to describe the capabilities of Microsoft Copilot Studio. This chapter introduces a tool that lets you build your own AI-powered chatbots and automated workflows — without needing to write any computer code. For PL-900, understanding what Copilot Studio does and how it fits into the Power Platform is essential because it represents the easy, low-code way to add conversational AI to business processes.
Jump to a section
A simple way to picture Introduction to Copilot Studio
A restaurant head chef’s recipe book is a collection of instructions for preparing dishes. The chef can follow the recipes exactly, step by step, to produce a consistent meal every time.
Now imagine that same chef has a new, clever apprentice called Copilot Studio. The apprentice doesn’t just follow the existing recipes — the apprentice can watch how customers order, notice that people often ask for a ‘spicy version’ of a popular pasta, and then suggest adding a new recipe to the book. The apprentice can also take spoken requests from waiters (like ‘a customer wants a gluten-free dessert that uses chocolate’) and turn those requests into a new, working recipe that the whole kitchen can follow.
The apprentice doesn’t cook the food itself. Instead, it designs the instructions so that other chefs (or even the restaurant’s automated equipment) can prepare the dish without needing to invent the method from scratch. If a waiter asks for ‘the lunch special for Tuesday’, the apprentice pulls the right recipe from the book and hands it over.
In this analogy, the recipe book is the set of automated workflows and chatbots inside your organisation. Copilot Studio is the tool that lets you create, edit, and improve those recipe-like conversational flows without needing a specialist software developer. Just as the apprentice understands kitchen language and produces usable recipes, Copilot Studio understands natural language and produces working conversational agents.
Microsoft Copilot Studio is a low-code tool inside the Microsoft Power Platform. ‘Low-code’ means that you can build functional software using a visual interface, dragging and dropping elements, rather than writing lines of programming language. Copilot Studio specialises in creating ‘conversational agents’ — also called chatbots or virtual agents. These are computer programs that can hold a text-based conversation with a person, answering questions, guiding them through tasks, or collecting information.
Before tools like Copilot Studio existed, building a chatbot required a team of software developers. They would write code to handle natural language processing (NLP) — the technology that allows a computer to understand human speech — and they would write complex logic to manage conversation flows. This was slow, expensive, and difficult to change.
Copilot Studio was created to change that. It gives you a graphical interface where you can design ‘topics’. A topic is a specific conversation route. For example, you might create a topic called ‘Check order status’. When a user types ‘Where is my parcel?’, Copilot Studio recognises that question and routes the conversation into the ‘Check order status’ topic. Inside that topic, you can add questions for the user (like ‘What is your order number?’), pull data from an external system (like a delivery tracking database), and then display the answer back to the user.
Copilot Studio uses a concept called ‘generative answers’. This means that for common questions, the bot can use AI to generate a response based on information you give it, like a website or a PDF document. You don't have to manually script every possible question. If a user asks something you haven't explicitly programmed, Copilot Studio can still try to answer using the content you provided.
Another key capability is ‘Power Automate integration’. Power Automate is another Power Platform tool for building automated workflows (a sequence of actions that happen automatically, like sending an email when a form is submitted). You can connect a Copilot Studio chatbot to a Power Automate flow so that, during a conversation, the bot triggers real-world actions. For example, a user could tell the bot ‘I want to request a leave of absence’, and the bot could run a flow that creates an approval request in Microsoft Teams.
Copilot Studio can be published to multiple channels. A ‘channel’ is a place where users interact with the bot. Common channels include a website, Microsoft Teams, Facebook Messenger, or a custom mobile app. You can publish your bot to one or many channels with just a few clicks.
Security is handled through ‘Microsoft Entra ID’ authentication. When a user signs in to the bot (for example, with their work email and password), the bot can identify who they are and only show them information they are permitted to see. This is called ‘role-based access control’ (RBAC) — different people get different views depending on their role in the organisation.
Copilot Studio also includes analytics. The product provides dashboards showing how many conversations the bot had, what topics were popular, where users dropped off, and how often the bot had to escalate to a human. An ‘escalation’ is when the bot cannot answer and passes the conversation to a real person. These analytics help you improve the bot over time.
In summary, Copilot Studio exists to democratise the creation of conversational AI. It replaces weeks of custom software development with a few hours of visual configuration. For business users and IT professionals alike, it means you can quickly build a helpful, secure, always-available assistant that handles routine questions and tasks, freeing up human workers for more complex problems.
Create a new chatbot in Copilot Studio
You sign in to the Copilot Studio web app and click 'Create'. You give the bot a name and choose its primary language (for example, English UK). This step establishes the identity of your bot — it is the container for all the topics, knowledge sources, and settings you will add later.
Add knowledge sources
You provide content that the bot can use to answer questions. This can be a link to a public website, a PDF file, or a SharePoint document. The bot scans this content and uses it to generate natural language answers when a user asks something you haven't explicitly scripted.
Create custom topics
A topic is a specific conversation path. For example, a topic called 'Check order status'. You define 'trigger phrases' — words or sentences that route the user to this topic (like 'where is my order'). Inside a topic, you add question nodes, message nodes, and condition nodes to build the conversation flow.
Connect to Power Automate flows
You create or select a Power Automate flow that performs a real-world action, such as looking up data in a database or sending an approval request. Inside the topic, you add a 'Call an action' node that runs this flow during the conversation, passing information from the user to the flow and back.
Test the chatbot
You use the built-in test pane in Copilot Studio to simulate a conversation. You type sample questions and check if the bot routes to the correct topics. You refine trigger phrases and adjust the conversation logic until the responses are accurate and helpful.
Publish the chatbot to channels
Once testing is complete, you publish the bot. You can choose one or more channels: a custom website (using a JavaScript snippet), Microsoft Teams, Facebook Messenger, or other platforms. Publishing makes the bot live and available to end users.
Monitor analytics and improve
After the bot is live, you review the analytics dashboard. You see metrics like total conversations, topic usage, escalation rate, and satisfaction scores. You use this data to identify weak points and edit topics to improve performance.
Consider a medium-sized retail company called ‘BrightHome Supplies’ that sells furniture and home goods both online and in physical stores. The company’s IT team consists of two generalists who support 200 employees. Customers frequently email the support team with simple questions like ‘What is your return policy?’, ‘Do you deliver to my postcode?’, and ‘Can I change my delivery date?’. The support team spends about 40 hours per week answering these repetitive queries.
An IT professional at BrightHome Supplies decides to use Copilot Studio to solve this. They do the following:
They open the Copilot Studio web app. They create a new chatbot and give it a name: ‘BrightHome Support Assistant’. They select the language (English UK) and the regional settings.
They upload the company’s return policy document (a PDF) and the delivery FAQ page (a web URL) as ‘knowledge sources’. This tells the bot to use these documents to generate answers automatically when customers ask related questions.
They create three custom topics. The first topic is ‘Check delivery date’. In this topic, they add a question node that asks for the order number. They use a Power Automate flow to look up the order in their cloud-based order management system. The flow returns the estimated delivery date, which the bot reads out to the customer. The second topic is ‘Return an item’. The bot asks what item, when it was purchased, and the reason. It then creates a return authorisation number and sends a confirmation email via another Power Automate flow. The third topic is ‘Talk to a human’. If the customer types ‘agent’ or ‘help’, this topic escalates the conversation to the human support queue in Microsoft Teams.
The IT professional tests the bot in the Copilot Studio test pane. They type sample questions and watch how the bot responds. They tweak the topics — for example, they notice the bot misunderstands ‘where is my stuff?’ so they add alternative trigger phrases like ‘order status’ and ‘delivery update’.
Once satisfied, they publish the bot to the company’s public website (using a JavaScript code snippet they copy from Copilot Studio) and also to the company’s internal Microsoft Teams tenant.
They then monitor the analytics dashboard. They see that the bot handles 80% of customer queries without human intervention. The support team’s workload drops significantly. They identify one topic where users frequently abandon the conversation (the ‘change address’ topic) and they rewrite it to make it simpler.
The IT professional in this scenario never wrote a single line of code. They used Copilot Studio’s visual interface, connected it to existing systems via Power Automate, and launched a working customer support chatbot in two days. What used to require a developer team and weeks of work is now achievable by a generalist with basic business process knowledge.
The PL-900 exam objective 5.1 asks you specifically to ‘Describe the capabilities of Microsoft Copilot Studio’. This is a descriptive question, not a deep technical one. The exam will not ask you to write code or configure advanced settings. Instead, you need to know what the tool can do, what it cannot do, and how it relates to other Power Platform components.
Here are the exact concepts the exam tests:
What is Copilot Studio? Look for the definition: a low-code tool for building conversational agents (chatbots) without writing code. The exam may offer a distractor like ‘a tool for building websites’ or ‘a tool for managing databases’. The correct answer will mention chatbots and conversational AI.
Key capabilities: You need to be able to list at least three. These include:
Creating topics to manage conversation flows.
Using generative answers from knowledge sources (PDFs, websites, SharePoint).
Integrating with Power Automate to trigger workflows.
Publishing to multiple channels (Teams, web, Facebook).
Analysing bot performance with built-in dashboards.
Authenticating users with Microsoft Entra ID for secure personalised experiences.
Common traps: The exam might show a scenario where a user wants to handle complex mathematical calculations. They might then ask which tool to use. Copilot Studio is not for heavy computation — it is for conversations. A distractor might be that Copilot Studio does its own machine learning training — it does not, it uses pre-built models. Another trap: they might say Copilot Studio requires a developer to write Python scripts — false, it is low-code.
Integration knowledge: Expect at least one question about connecting Copilot Studio to Power Automate. Remember the pattern: a user asks the bot to do something (like approve an expense), the bot runs a Power Automate flow that performs the action.
Channels question: You may be asked which platforms a Copilot Studio chatbot can be published to. The correct answer includes Microsoft Teams, a custom website, and Facebook Messenger. A common distractor is ‘email’ — you cannot publish a chatbot directly in an email (though you can send email via Power Automate flows triggered by the bot).
Authentication concept: Know that Microsoft Entra ID is used for single sign-on (SSO) so that users do not have to log in separately to the bot. The bot can then show personalised information based on who the logged-in user is.
The exam will not test: Specific versions, pricing tiers, API calls, or Azure configuration details. Stick to the high-level capabilities as described in Microsoft’s official documentation.
To summarise, for a PL-900 question on Copilot Studio, your answer should always reference its low-code nature, its purpose (building chatbots), its integration with Power Automate, its ability to publish to multiple channels, and its analytics features. If the question asks what problem it solves, say it enables non-developers to create conversational AI agents quickly.
Copilot Studio is a low-code tool for building conversational agents (chatbots) without writing programming code.
It integrates with Power Automate to let your chatbot trigger workflows like sending emails or creating approval requests.
You can publish a Copilot Studio chatbot to multiple channels, including Microsoft Teams, websites, and Facebook Messenger.
Copilot Studio uses generative answers to respond to user questions from knowledge sources like PDFs, web pages, and SharePoint documents.
Authentication with Microsoft Entra ID lets the chatbot recognise users and show personalised, role-specific information.
Built-in analytics in Copilot Studio show you how many conversations the bot handled, what topics were used, and where users needed to talk to a human.
Topics in Copilot Studio are pre-defined conversation paths that you design to handle specific questions or tasks.
Creating a chatbot in Copilot Studio does not require a developer — business analysts and power users can build one using a visual interface.
These come up on the exam all the time. Here's how to tell them apart.
Copilot Studio
Used to build custom chatbots through a low-code interface.
Can be published to multiple channels like Teams, web, and Facebook.
Integrates with Power Automate for custom workflows.
Microsoft Copilot (M365)
Built into Microsoft 365 apps like Word and Excel to assist with content creation.
Only works inside Microsoft 365 apps.
Cannot be customised with external workflows or databases.
Copilot Studio (low-code chatbot builder)
Requires no programming skills; uses visual drag-and-drop interface.
Designed for business analysts and power users.
Handles natural language processing automatically behind the scenes.
Microsoft Bot Framework (code-based chatbot builder)
Requires coding in C# or JavaScript.
Designed for professional software developers.
Developer must implement natural language processing logic manually.
Custom topic in Copilot Studio
Manually created conversation path with specific questions and logic.
Used for known, repetitive scenarios like 'check order status'.
Gives you full control over the conversation flow.
Generative answer in Copilot Studio
AI-generated response based on provided knowledge sources.
Used for unexpected or varied questions not covered by custom topics.
Less control; the AI decides how to phrase the response.
Copilot Studio analytics
Built-in dashboards specifically for chatbot performance (conversations, topics, escalations).
Available directly inside Copilot Studio without extra setup.
Best for real-time monitoring of your bot's health.
Power BI dashboards
General-purpose business intelligence tool for any data source.
Requires a separate Power BI license and data integration.
Best for complex, cross-system reporting and visualisations.
Mistake
Copilot Studio requires programming skills to build a chatbot.
Correct
Copilot Studio is a no-code/low-code tool. You build chatbots using a visual interface with drag-and-drop elements, not by writing code.
Beginners often assume that any AI tool must require software development. They confuse Copilot Studio with platforms like Microsoft Bot Framework that do require coding.
Mistake
Copilot Studio is the same as Microsoft Copilot (the AI assistant in Office apps).
Correct
Microsoft Copilot (in Microsoft 365) helps you write documents and emails. Copilot Studio lets you build your own custom chatbots. They are different products.
The name 'Copilot' appears in multiple Microsoft products. Beginners mix them up because they sound similar, but their purposes are completely different.
Mistake
If a user asks a question the bot hasn't been explicitly taught, the bot will just fail or give an error.
Correct
Copilot Studio can use generative answers. If it hasn't been taught a specific topic, it can still generate a response from knowledge sources like a PDF or a website you provide.
People expect traditional chatbots that need every script written in advance. They don't realise that modern AI can answer novel questions using provided content.
Mistake
Copilot Studio can only be used by IT professionals with special permissions.
Correct
Copilot Studio is designed for all types of users, including business analysts and power users. Anyone with a Power Platform license and appropriate permissions can create a bot.
There is a common belief that powerful tools are locked away for IT only. Microsoft markets Copilot Studio as a 'citizen developer' tool, meaning business users can build things themselves.
Reveal each answer, then mark whether you got it right. Score 60%+ to unlock the next chapter.
No. Copilot Studio is a low-code tool. You build chatbots using a visual interface with drag-and-drop elements, so you do not need to write any programming code.
Yes. If you provide knowledge sources like a PDF or a website, Copilot Studio can use generative AI to answer questions based on that content, even if you haven't created a specific topic for that question.
You can publish to several channels including Microsoft Teams, a custom website, Facebook Messenger, and other supported platforms. You cannot publish directly inside an email.
No. Microsoft Copilot in Word helps you write documents. Copilot Studio is a separate product for building your own custom chatbots. They share the 'Copilot' name but have different functions.
Copilot Studio can use Microsoft Entra ID for authentication. When a user signs in, the bot knows who they are and can show information based on their role or permissions using role-based access control (RBAC).
Yes. You can connect Copilot Studio to other systems using Power Automate flows. For example, a flow can look up a customer order from a database and the bot can read the result to the user.
You've finished Introduction to Copilot Studio. Continue through the PL-900 study guide to build a complete picture of the exam.
Done with this chapter?