Exercise 2, Task 3: Use Copilot Studio to build an RFP response agent

While Fabrikam’s sales success with EcoSense 360 generated strong interest from hotels and resorts, it also created a new challenge. The Sales team is spending hours each week responding to early-stage RFPs that ask similar questions about integrations, energy savings, and product capabilities. To address this bottleneck, Fabrikam’s VP of Sales tasked you with developing a Copilot agent that can automatically handle these initial inquiries.

You plan to use Copilot Studio to build an agent that performs two functions: first, it provides preliminary answers drawn from existing EcoSense 360 product materials, and second, it generates a response for a submitted RFP. This agent should reduce response times, free the Sales team to focus on higher-value opportunities, and give prospective clients a faster, more engaging experience.

[!NOTE] In this exercise, you use the Copilot Studio lite experience to create the EcoSense 360 RFP Response Agent. This simplified experience is designed for everyday business users and requires no programming skills. By contrast, software developers who build more complex, advanced agents typically use the full Copilot Studio experience.

Perform the following steps to complete this task:

  1. Select the following links to download all the documents that will provide knowledge source content for the new EcoSense 360 RFP Response Agent:

    Store these files in the EcoSense360-RFP-Documents folder that you created in your OneDrive in Task 1. These documents provide the knowledge sources for the EcoSense 360 RFP Response Agent.

  2. Select the following link to download the RFP document that you plan to use when testing the new EcoSense 360 RFP Response Agent:

    Store this RFP file in your OneDrive folder.

[!CAUTION] Don’t store the VanArsdel_RFP.docx file in the EcoSense360-RFP-Documents folder. This folder is only for the knowledge source documents for the new agent.

  1. Open a new tab in your Microsoft Edge browser and then open Microsoft 365.

  2. In Microsoft 365, select New agent in the navigation pane. Doing so opens Copilot Studio’s Agent Builder and displays the New agent page.

  3. On the New Agent page, you want to ask Copilot to create an agent. In the prompt, you should enter the agent’s name and a general description of what the agent is about, who its target audience is, and what you want it to do.

    For this agent, enter the following prompt and then select the forward arrow (Send) icon to submit the prompt:

    Create an agent titled EcoSense 360 RFP Response Agent. The purpose of this agent is to provide the Sales team with answers to questions related to customer sales proposals. It should also be able to generate a complete, customer-ready response to customer RFPs for the EcoSense 360 energy management solution. For knowledge sources, the agent should only use the documents stored in the EcoSense360-RFP-Documents folder. RFP responses should be based on the template file titled: EcoSense_360_RFP_Template.dotx.

  4. After you selected the forward arrow, the Agent Builder form appeared for your new agent. At the top of the form is a Describe tab and a Configure tab.
    • The Describe tab enables you to carry on a conversation with Copilot. This tab is displayed by default.
    • The Configure tab enables you to define the detailed settings that drive the agent.

    Wait a minute or two for Copilot to create the agent, at which time it displays the agent’s name and description in the Agent preview pane.

  5. Select the Configure tab at the top of the form. Let’s see what Copilot did based on the prompt that you entered.

  6. On the Configure tab, the Name and Description fields should be filled in based on the prompt that you entered. Scroll down to the Instructions field. Copilot generated these instructions based on the description that you provided in your initial prompt. Review the detailed level of instructions that Copilot generated.

[!IMPORTANT] The beauty of the Agent Builder process is that Copilot automatically translates your basic, natural language description into a complex set of instructions. This process saves you from creating this detailed instruction set on your own.

  1. If you wish to change the instructions, you can either manually edit them directly in the Instructions field, or you can ask Copilot to update the instructions for you.

    After reviewing the Instructions, you decide that you want to have Copilot add a couple of other items to the instruction set. To do so, select the Describe tab and then enter the following prompt:

    Update the Instructions to include the following items:

    • Responses must follow the RFP’s formatting rules, include tables for compliance and pricing, and maintain a professional, customer-focused tone.
    • Ensure responses are formatted for clarity with short labels and bullet points where appropriate.
    • Cite sources for every answer.
    • Flag missing information.

  2. Review Copilot’s response after updating the instructions. To verify the changes that Copilot made, select the Configure tab and then scroll down to the Instructions field. Verify that Copilot added the new instructions that you requested.

  3. While the current instructions look good, you wonder if they could be improved upon. You aren’t sure how to improve them, so you decide to ask Copilot what it thinks.

    To do so, select the Describe tab. This time, enter a prompt that asks Copilot what other instructions it would recommend that could improve this agent.

  4. Review Copilot’s recommendations. You’re pleased with its suggestions, so ask Copilot to add them all to the agent’s instructions.

  5. Once Copilot responds that it updated the instructions, select the Configure tab and scroll through the Instructions. Note the new items that Copilot added.

  6. Now that you’re satisfied with the instructions, you’re ready to configure the agent’s knowledge sources and starter prompts.

    In the Configure tab, scroll down to the Knowledge section and verify the Search all websites toggle switch is disabled. Copilot should have disabled this toggle switch when it created the agent based on the description you provided in your original prompt, which told it to only use the files stored in the EcoSense360-RFP-Documents folder. If the toggle switch is enabled, then disable it now.

  7. In the Knowledge section, select the Upload from device icon that appears next to the Enter a URL or name or drop files here field. In the File Explorer window that appears, navigate to your OneDrive folder and select the EcoSense360-RFP-Documents folder. Select all of the files in this folder and then select the Open button.

  8. Scroll to the bottom of the Knowledge section and verify all the selected documents appear in the list of Uploaded files.

  9. For Suggested prompts, you can have Copilot generate prompts for you, or you can manually create your own prompts. Let’s try both methods.

    To have Copilot generate suggested prompts, select the Describe tab and then ask Copilot to generate three suggested prompts for the agent. Note how each prompt has a title and a message.

  10. You now want to enter several of your own prompts. Select the Configure tab and scroll down to the Suggested prompts section. You should see the three prompts that Copilot added to the agent.

    For each prompt that you want to manually add, select the Add a suggested prompt option that appears below the prompts.

    Six suggested prompts are displayed below that ask questions about the EcoSense 360 energy management solution. Review these prompts, select two or three that you like, and then add them to the agent.

    • Title: Product overview and benefits
      • Message: Provide a product overview and list of benefits.
    • Title: Competitor differentiation
      • Message: Provide a one-page summary of differentiators compared to competitors.
    • Title: Product integration
      • Message: Explain how the product integrates with existing hotel management systems.
    • Title: Product integration
      • Message: How does the EcoSense 360 integrate with HVAC and lighting systems?
    • Title: Pricing
      • Message: What pricing options are available for large resort chains?
    • Title: Energy savings and ROI
      • Message: Draft an executive summary highlighting energy savings and ROI for hotels.

  11. Test several of the suggested prompts. Verify the agent is correctly pulling in data from the knowledge source documents.

  12. You now want to add a final suggested prompt that asks the agent to generate an RFP document based on an attached sales proposal file.

[!NOTE] We held off on having you enter this suggested prompt until after you finished testing some of the prompts in the prior step. We did so because the Test pane in Copilot Studio currently doesn’t let you attach files in its prompt field. As such, we wanted to avoid any confusion when you tested the prompts in the prior step.

Add the following suggested prompt (which you test in later steps):

  • Title: Generate an RFP response
    • Message: Create an RFP response for the attached sales proposal. Format the result into a Word document based on the EcoSense_360_RFP_Template. Provide a link to download the document.

  1. The agent’s configuration is now complete, so select the Create button to create the agent.

  2. Once the agent is created, a dialog box appears that indicates the agent was successfully created. In this dialog box, you can either go to the agent or share it. Select the Go to agent option.

[!NOTE] At this stage, the agent is private and accessible only to you. In a real-world scenario where the agent needs to be used by multiple team members, you would share it with those individuals. For this training exercise, sharing isn’t required since you’re working within your own tenant.

  1. You now want to test the final suggested prompt that you added earlier, which creates an RFP response to a sales proposal.

    In the EcoSense 360 RFP Response Agent window, select the Generate an RFP response suggested prompt (Create an RFP response for the attached RFP). Attach the VanArsdel_RFP.docx file from your OneDrive account and then submit the prompt.

  2. Review the results. One of two results might occur:

    • During our testing, Copilot indicated that it couldn’t generate a downloadable Word document. If this situation happens to you, then select the Copy option at the end of the results to copy the RFP response to your clipboard. Then open Word, paste in the copied text, delete any extraneous verbiage that might appear at the start and end of the results, and then save the file to your OneDrive as VanArsdel-RFP-Proposal.docx. Leave the file open.

    • Since Microsoft 365 Copilot is still a work in progress, it might provide a downloadable Word document for you by the time you perform this exercise. If Copilot generates the document, select the link to download it, and then open the downloaded Word document.

[!IMPORTANT] Before creating the Word document or downloading the generated document (if it’s able to do so), review the suggested prompts at the end of the results. Feel free to submit any of these prompts if you want the agent to update the results. For example, it might ask whether you want to include a cover page and table of contents. Or, it might ask if you want it to make more updates, such as expanding the compliance matrix with more requirements. Now is your opportunity to have the agent customize the RFP response with any extra features.

  1. Review the RFP response in the Word document. Remember, the agent used the predefined template file titled EcoSense_360_RFP_Template.dotx as the basis for its response. If it didn’t fill out any of the fields in the template, then you must manually update them yourself (or remove or replace them).

    At this point, you can still use the agent to help you. If you want, you can optionally submit a prompt to the agent that contains your query and then copy and paste the result into the RFP response file where appropriate.

  2. Leave the VanArsdel-RFP-Proposal.docx open as it’s used in the final task in this exercise.