Reading-Discussion Chatbot Builder
UO Online
A UO Online module

Reading-Discussion Chatbot Builder

A chatbot is an AI program that lets students discuss a reading with it and test their ideas, interpretations, and questions. It responds with questions that push them to explain, support, and reconsider their reasoning. In about 10–15 minutes, you'll set one up for your course so that it prompts students to think for themselves rather than handing them answers. No AI experience needed.

About 10–15 minutes

What you'll need

  • Your reading(s) as a file or a link. Have more than one if you want students to compare texts.
  • Your UO login for Microsoft Copilot (your UO email and password).
  • Your Canvas course, where you'll add the link to your finished chatbot.

New to AI? anytime.

Build a prompt for your reading-discussion chatbot

This page asks you a few simple questions, then turns your answers into one finished piece of text called a "prompt" — instructions that tell an AI tool how to act. When you're done, you'll paste that text into an AI tool to create your chatbot.

This page doesn't save anything — if you close it partway through, you'll need to start over, so once your prompt is finished, copy it somewhere safe (like a document on your computer).

This page is built around Microsoft Copilot. UO has an agreement that gives every faculty member and student free, secure access at copilot.cloud.microsoft — that's why it's the recommended path. Either way, the text you build here will work anywhere: you can also paste it into Claude, ChatGPT, or another AI website instead.
Have your reading ready before you finish this process. Near the end, you'll give your chatbot the actual reading — as a file or a link — directly in your AI platform. That's what lets the chatbot discuss your specific reading and sources with students, so it's worth keeping them nearby. You don't need them yet, though — feel free to start building your prompt below.

This page was built with Claude, Anthropic's AI assistant.

Step 1 of 5  ·  Guide Step 1

What's the goal?

Before anything else, it helps to decide what you want students to get out of talking with this chatbot. Just answer these in your own words — there's no wrong way to fill this out.

Step 2 of 5  ·  Guide Step 4

Choose a personality and conversation style

These questions shape how the chatbot "sounds" and behaves. Pick whatever feels right — you can select more than one, or write your own.

This is about the chatbot's approach during the conversation itself — not its personality, but its technique. Getting this right is what keeps a conversation feeling like a real dialogue instead of a lecture or a quiz.

Step 3 of 5  ·  Guide Step 5

Guide the discussion

Add any specific questions, themes, or expectations you want the chatbot to bring into the conversation. You'll give it the actual reading itself later, in Step 5 — this step is just about how you want the discussion shaped. If you're working with more than one text, this is also where you can describe what you want the chatbot to help students compare, connect, or contrast across them.

Step 4 of 5  ·  Guide Step 6

Review and generate your prompt

One last question, then this page will assemble everything above into one finished prompt.

Generate your prompt above to continue.
Step 5 of 5  ·  Guide Step 7

Build your chatbot and add it to Canvas

You should have your prompt copied or downloaded from the last step. Canvas doesn't build the chatbot itself — you'll build it on an AI platform, then add a link to it in Canvas so students can reach it. First, which platform are you using?

Copy it now, or expand to keep it visible while you follow the steps below.

Follow these steps

Complete these steps now, in Microsoft Copilot, to turn your prompt into a working chatbot and get it in front of students.

  1. Go to copilot.cloud.microsoft and sign in with your UO email and password. Look for the green shield icon — that confirms you're in the protected, UO-approved version. Open Agent Builder and create a new agent.
  2. Paste your prompt into the agent's instructions. Then, on that same setup screen, scroll to the "Knowledge" section. If a reading is a file, click "Upload" (or just drag it in); if it's a web page or online article, click "Enter URL" and paste the link there instead. You can add more than one — mix files and links freely if you're using several texts. This is what actually gives your chatbot the full reading(s) to work with — more on why below.
  3. Share the agent with "Specific people in your organization," then copy its link and add that link into Canvas (for example, as a module item) so students can reach it.

Why attach the reading(s), too? The prompt text alone can't include your full reading(s) or a live link to them — attaching the actual file(s), or entering their URL(s), as "knowledge sources" gives your chatbot everything it needs to work with.

Not sure which to use? If your campus hasn't pointed you toward something else, Microsoft Copilot is the safest default — it's free for every UO student and already approved for this kind of use. Ask your instructional designer or IT support if you're unsure.

You've got your prompt. Here's what's left.

These steps don't change the text you just built, but they matter for actually getting your chatbot in front of students. They're drawn from your full guide, Building an AI Reading-Discussion Chatbot.

Before you launch

Getting your course ready

After launch