--- title: "Create Your Own Copilot Agent: A Personalized AI Assistant for Teaching" type: "blog" space: "The Teaching & Learning Centre" author: "Ashley Ribbel" created: 2026-10-23T16:21:34Z updated: 2026-10-28T13:48:37Z tags: ["ai", "artificial-intelligence", "spark-plug", "teaching", "teaching-and-learning", "teaching-and-learning-centre"] url: "https://employees.senecapolytechnic.ca/spaces/39/the-teaching-learning-centre/blogs/updates/24878/create-your-own-copilot-agent-a-personalized-ai-assistant-for-teaching" --- # Create Your Own Copilot Agent: A Personalized AI Assistant for Teaching [**View all the Teaching & Learning Centre Spark Plugs!**](https://employees.senecapolytechnic.ca/spaces/39/the-teaching-learning-centre/blogs) * * * **by Iman Hajihassani, Multi-Media Specialist in the Teaching & Learning Centre** **If you’ve used Microsoft Copilot before, you’ve probably experienced this: you open a new conversation and begin explaining everything all over again.** “I teach this course.” “My students are at this level.” “This is how I like my announcements written.” “Please keep the language clear, friendly, and concise.” Then, a few days later, you need help with something else and start the process again. But what if you could create a version of Copilot that already knows what kind of help you want? That’s where Copilot agents come in. ### What is a Copilot agent? A Copilot agent is a personalized AI assistant that you give a specific role and instructions. Instead of starting with a completely blank conversation every time, you can create an agent designed around a particular teaching task or workflow. The useful part is that you don’t need to be a programmer to create a basic agent. You can describe what you want it to do in everyday language, then refine its instructions as you use it. ### Start with one teaching job Imagine creating an agent called My Course Assistant. Its job could be to help with weekly announcements, discussion questions, activity ideas, and clear explanations for students. Rather than trying to make one agent do everything, start with a task or group of tasks you repeat often. > > > **Try this as your starting instruction:** > > > You are an assistant for a college professor. Help me create clear weekly announcements, discussion questions, activity ideas, and explanations for students. Keep the tone supportive and conversational. When suggesting activities, focus on active learning and provide options for both online and in-person classes. > You can adjust that instruction for your own course, teaching style, and priorities. The more clearly you describe the agent’s role, audience, tone, and boundaries, the more consistently it can help. ### Create your first agent On Seneca’s Copilot desktop app, select Agents from the left-hand menu. From the Agents page, choose New agent. ![Step 1. Open Agents and select New agent.](https://employees.senecapolytechnic.ca/attachment?file=SwasWOYUXj98PhnX6Yme2A%3D%3D) Step 1. Open Agents and select New agent. Give the agent a clear name and describe how you want it to help. You can start with a simple set of instructions like the example above, then improve them later as you discover what works well. ![Step 2. Give your agent a name, add its instructions, and select Create.](https://employees.senecapolytechnic.ca/attachment?file=R2xpGaePmt%2FdA4tA3lzfQw%3D%3D) Step 2. Give your agent a name, add its instructions, and select Create. ### Use it like a teaching assistant Once the agent is created, you can ask it the kinds of questions you deal with during a normal teaching week. For example: - Help me create an activity for next week. - Turn this topic into five discussion questions. - Draft a friendly announcement reminding students about the assignment. - Give me three different ways to explain this concept to students who are struggling. ![Step 3. Test the agent with a real teaching task and review the response.](https://employees.senecapolytechnic.ca/attachment?file=FUYIa4oohdMG0w0FojJX8w%3D%3D) Step 3. Test the agent with a real teaching task and review the response. If the answer is too long, tell the agent to be shorter. If the tone is too formal, update the instructions. If you want it to always give several options instead of one, add that expectation. Think of the first version as a draft: you’re teaching the assistant how you want to be helped. ### Build around the work you repeat Different agents can support different parts of teaching. One might focus on active-learning ideas, another on assessment brainstorming, and another on turning complex concepts into plain language. Keeping each agent focused can make its purpose clearer and its responses easier to evaluate. As with any AI-generated content, you still review the output, make the final decisions, and ensure that the information and materials you provide are appropriate to share. The goal isn’t to hand teaching over to AI. It’s to reduce repetitive work so you can spend more time on the parts of teaching that need your experience and judgement. ### Start small, then build For now, try creating one small agent around a task you find yourself doing again and again. Test it with a few real examples and refine the instructions until the results feel useful. In a future Midday Mindshare: AI for Teaching workshop in the AI Lab, we’ll go further: improving agent instructions, adding useful knowledge, and exploring how agents can support recurring teaching workflows. **You might be surprised by how much more useful Copilot becomes once it knows what job you want it to do.** * * * [**View all the Teaching & Learning Centre Spark Plugs!**](https://employees.senecapolytechnic.ca/spaces/39/the-teaching-learning-centre/blogs) * * * ## Attachments - [Picture1.png](https://employees.senecapolytechnic.ca/attachment?file=SwasWOYUXj98PhnX6Yme2A%253D%253D) - [Picture2.png](https://employees.senecapolytechnic.ca/attachment?file=R2xpGaePmt%252FdA4tA3lzfQw%253D%253D) - [Picture3.png](https://employees.senecapolytechnic.ca/attachment?file=FUYIa4oohdMG0w0FojJX8w%253D%253D)