Ready to tap into the limitless opportunities offered by agentic AI? Microsoft Copilot Studio is the low-/no-code toolkit created by Microsoft that empowers anyone to build, customize, and deploy autonomous AI agents.
Introduced in November 2023 at the Ignite conference, Copilot Studio first enabled businesses to create custom use cases for a human-in-the-loop virtual assistant.
Yet, it’s fast becoming a hub for “human at the helm “AI agents that cross enterprise platforms, leveraging various generative AI foundation models and secure business data to autonomously complete tasks.
But that’s the whistle-stop tour explanation. Below is a more comprehensive guide to the ever-evolving Copilot Studio.
What Is Microsoft Copilot Studio?
Microsoft Copilot Studio is a comprehensive low-code toolkit that allows users to both create their own custom agentic AI (autonomous AI) bots and customize pre-built AI agents.
With the Studio, anyone can design agents using a low-/no-code interface (or natural language prompts).
They can connect those agents to their data sources using pre-built or custom plugins and publish for various environments, including Microsoft 365 apps.
The toolkit builds on Microsoft’s Power Virtual Agents tool, designed for the era of generative and autonomous AI.
It’s integrated with the Power Platform and Power Platform admin center and is available worldwide – even in Government Community Clouds.
When it was first released, Copilot Studio primarily enabled the creation of custom “GPT” bots – similar to Microsoft’s Copilot tools for solutions like Microsoft Teams or ChatGPT.
At Ignite 2024, however, Microsoft upgraded Copilot Studio with new agentic AI features. That means you can now create “Agents” that can complete multi-stage tasks autonomously.
These agents coordinate language models, instructions, context, actions, topics, inputs, knowledge sources, and triggers to accomplish specific goals.
Key features of Copilot Studio include:
- Agent Development or Customization: Either build a new agentic AI model from scratch or customize one of Microsoft’s pre-built agents with your internal knowledge from your assets, websites, Microsoft Graph, and the Microsoft Dataverse.
- AI Deployment: Publish agents directly to the applications your employees already use, such as your website, Microsoft Teams, SharePoint, social media, and more.
- AI Management and Governance: Streamline AI governance with guidelines, access controls, and visibility into agent actions through Power Platform.
Microsoft Copilot Studio: Agent Use Cases
The potential for autonomous AI agents is practically limitless – particularly in the worlds of sales, marketing, and customer service.
A complete list of potential use cases for agentic AI solutions is available here, but just some examples of what businesses can build with Microsoft Copilot Studio include:
- Retail bots: Intelligent agents that can proactively minimize abandoned carts with custom, personalized offers, issue supply alerts, and deliver post-purchase assistance.
- Hospitality bots: Copilots that can manage flights and schedules in real-time, create personalized itineraries, and handle customer service requests.
- Customer service bots: Agents that can handle everything from order update insights to purchasing assistance, troubleshooting, and onboarding.
- Employee copilots: Intelligent assistants that can handle time off requests, deliver training to team members, or answer common questions about tasks and products.
- Sales bots: Bots capable of creating sales pitches for agents, personalizing product recommendations, and tracking inventory details.
How Much Does Microsoft Copilot Studio Cost?
The great thing about Microsoft Copilot Studio is how easy it is to use – even if you don’t have any programming or data science knowledge. Yet, to get started, the first thing a business will need is a license for Copilot Studio.
While a license for Microsoft 365 Copilot costs $30 per month, per user, Copilot Studio costs much more.
Indeed, Microsoft Copilot Studio is currently available at $200 per month, per user: This standalone plan allows users to build agents available across multiple channels for customer and employee assistance.
Yet, there’s also a pay-as-you-go plan that allows users to pay exclusively for what they build and use. This plan requires an Azure subscription too.
Once a brand has a plan, users can access Copilot Studio either through the standalone web app or through Microsoft Teams.
How to Create a Custom Agent in the Copilot Studio Web App
When it comes to creating an agent, Microsoft offers the option to either “build” a new agentic AI bot from scratch or extend the existing Microsoft 365 Copilot (or another pre-built agent) with new features.
Users can customize an existing agent by simply logging into their Studio account, selecting the “Agents” panel, and choosing “Microsoft 365 Copilot”.
From there, they can select the “Add” button, and they’ll be sent to a conversational authoring experience. There, they can describe what they want their agent to know and do.
Microsoft’s guide here showcases how to “add” knowledge to the customized agent using various plugins and tools.
To create a Copilot Agent from scratch using the web app, a user can:
- Log into Copilot Studio with their Microsoft credentials, and click “Create” on the left-hand panel. Select “New Agent” when the option appears.
- Use the chat to describe what the agent should do. A tip here is to keep the instructions simple and structured. Users can also describe the tone they want for the agent.
- Add an image to represent the agent by clicking the agent icon in the top bar and select “Change Icon”. Then, choose an image (PNG format).
Optimizing and Publishing an AI Agent
After creating an agent, a user can customize it further.
To do so, select “Agents” in the left navigation panel within Microsoft Copilot Studio, then choose the individual agent.
From there, click the “Topics” tab to make changes to certain things, like how an agent introduces itself to a user. To change basic elements:
- Click on the “Overview” tab and select “Details”
- Edit details such as the agent’s name, description, or instructions, or scroll to the “Knowledge” card to click “Add Knowledge”, and infuse the agent with extra data.
- Select the type of knowledge that the agent should use, such as public websites or specific file locations within theMicrosoft ecosystem.
Additionally, users may test their agents by asking questions and selecting the “Details” card to edit agent instructions.
From there, the user can publish it to a website or another source by selecting “Publish” at the top of the page.
How to Create a Custom Agent in Microsoft Teams
Creating an agent with Microsoft Copilot Studio within Microsoft Teams is pretty straightforward too. All the user needs is the Copilot Studio integration for Microsoft Teams and the right license.
To access the app, go to the Teams app store and search for “Copilot Studio” then click “Add” or “Open”. Users will then be taken to the Copilot Studio home page.
From there, they can click “Start Now” and choose the team they want to manage the agent. Then, they can give the bot a name (describing what it’s going to do, such as HR agent).
Finally, pick the language that the agent should communicate in, and select an icon, then hit “Create”.
Customizing an AI Agent on Teama
Each agent created in Teams will come with a collection of “topics” that it can provide support with. Click on the “Topics” tab to see the topics available, then select “New Topic” to create a new one.
Select “From Blank” and choose the “Details” option to adjust the topic details. Users can add trigger phrases like “Information on Time Off” for a bot designed to help with HR topics.
From here, they’ll be able to create “question flows” that identify the kinds of responses agents will give to each “prompt” chosen.
Copilot Studio users can also leverage “Variable properties” to allow the agent to dynamically respond to different types of questions – like questions about paid and unpaid time off.
Plus, there are options to adjust how the agent ends a conversation, like telling it to send a survey to end-users to see how helpful the agents are.
Microsoft recommends testing each agent with the “Test Agent Canvas” to ensure it’s responding appropriately to questions before publishing.
Once ready to publish your agent, users can hit “Publish” in the navigation menu. They can also “Open the Agent” on this page and select “Share the bot” to ensure it’s available to employees in the Teams app store.
Once the agent is published, it should be available in the “Agent Extensions” section of the Teams app store, and anyone in the company will be able to add the agent to Microsoft Teams.
Getting Started with Microsoft Copilot Studio
Microsoft Copilot Studio makes it simple for any organization to take advantage of the latest innovations in agentic AI. Whether you’re looking for a bot that can help you handle customer queries, streamline sales strategies, or even improve the employee experience, it only takes minutes to build an autonomous agent with this toolkit.
Want to learn more about the opportunities offered by agentic AI technology? Check out CX Today’s proprietary report on 2025: The age of AI Agents here.