For the past couple of years, I have talked about wanting a chatbot. In the client websites I am overseeing, in the automatic support responses, in the courses I am designing and teaching. For the past couple of years, building such chatbots always moved to the backburner. I didn’t have the time, I didn’t have the skills, I didn’t have the resources to hire people who had the time and the skills.
This week, I built two knowledge-specific AI chatbots. One for a website. One for a course.
One is designed to prescreen qualified leads. The other is a course coach. I have a bit more work to do training them… but the results are already astounding. And it took me two days.
Two days.
Not two months. Not a funded innovation grant. Not a cross-functional committee. Two focused days.
The Real Shift: From “Someday” to “Now”
For years, the idea of a course chatbot felt like an IT project. Something that required a developer. Budget approval. A strategic roadmap.
Now? If you can design a course, you can design a course coach.
What changed is not just the tools. It’s the usability of the tools. AI models can now be trained on your specific materials — your syllabus, your frameworks, your assignments, your FAQs.
That means we can stop talking about AI in the abstract and start building specific, contained, pedagogically aligned supports.
What a Course Coach Actually Does
Let’s be clear: I’m not talking about replacing faculty. I’m not talking about automating grading. I’m not talking about outsourcing intellectual labor.
I’m talking about structured support between class meetings.
The course chatbot I built is trained on specific course materials. It functions as:
A reflection partner
A clarifier of expectations
A guide through structured exercises
A prompt generator for deeper thinking
A first-pass support before a student emails me
Students can ask:
“What does this assignment actually require?”
“How do I apply this framework to my own situation?”
“Can you walk me through the first step?”
“I’m stuck — where should I start?”
It responds based only on the knowledge base I provided.
This is not a general AI wandering the internet. It’s a bounded, knowledge-specific coach.
Demo
Let’s start with the product, so it’s clear what I’m talking about.
This is an AI course coach built for ChangePRO, the personal change management workshop I facilitate.
It’s trained specifically on the workshop’s frameworks, exercises, and materials. It’s bounded. It stays within scope. It reinforces the methodology rather than improvising beyond it.
ChangePRO helps professionals navigate change intentionally — clarifying goals, addressing resistance, and building actionable plans. The AI coach simply extends that learning between sessions.
That’s it.
A course-embedded AI coach, aligned to the curriculum.
And yes, I built it in a weekend.
What you see here is built on Zapier but it can be embedded on any platform that has an HTML code editor and allows running scripts or embedding an iFrame. So far, I have tested it in WordPress, LearnWorlds and Canvas.
Interested in building your own?
Here is a step-by-step guide.
What is Zapier?
In a nutshell, Zapier is a no-code automation platform that connects different applications so they can exchange information automatically. Users create workflows called “Zaps,” where a trigger in one system prompts an action in another — for example, sending data from one tool to another without manual intervention. Zapier is a great tool that deserves a separate conversation. Let’s set that aside for now though and head straight for the Chatbot feature.
⚠️Disclaimer! If you decide you want to create a chatbot and embed it in your course, you will need a PAID account and it will run you about $160 USD annually for five chatbots.
Ready? Let's Do This!
Create a New Chatbot
Go ahead and create your first chatbot by clicking on Create from scratch.
Give it a name, such as Demo Chatbot, and click on Create.
Directive
The two most important things in setting up these chatbots are Directive and Knowledge. This is pretty much what sets this chatbot apart from the good old Chatty GPT. Directive here is basically instructions on how the chatbot should respond in terms of goals, style, etc. If the chatbot is going to act as a Teaching Assistant, tell it precisely that. This is where you can also tell it what to do (or what absolutely not to do) in specific circumstances. This is where you can use ChatGPT to help you write a comprehensive directive.
Knowledge
This is the other crucial piece of the puzzle. Unlike ChatGPT and other Ai platforms that will search the internet for every random piece of information and make stuff up on the spot, Knowledge limits and focuses out chatbot on the specific sources.
If you are new to Zapier and are using one of the lower subscription tiers, your knowledge sources are limited to websites and text-based files. I have also found that it’s more efficient (and will make the bot work faster) to make a text-based summary of different sources than to give it multiple links. This is where you can upload syllabi, course documents, assignments, assessments, presentation transcripts, and whatever else you can think of.
Again, I used ChatGPT to compile summaries of peer-reviewed articles, seminal studies, and other sources used in the workshop, in several text-based documents. Depending how you structure these documents, you can train your chatbot to suggest specific materials for review.
Once your knowledge sources have been uploaded, you will see the option to adjust Knowledge Settings.
Otherwise, you are pretty much done! You can explore other options and settings, but, for the most part, you are ready to embed your chatbot in the course.
Embed in the Course
Zapier allows several ways to embed chatbots but not all systems will allow all methods. The simplest and most reliable method is to use <iFrame> tags in the LMS “lessons”. If you have ever embedded a YouTube video, this is very similar.
In your chatbot view in Zapier, click on Share in the upper right corner.
Click on Embed –> Inline –> iframe. Copy code.
Now open your LMS course, create a lesson (page), switch to the code view, paste the code and save.
Here it is in Canvas in code view.
And here it is back in page view.
Conclusion and Next Steps
Building a course-aligned AI coach is no longer a massive technical project. It’s doable. And it can meaningfully support student learning when it’s aligned to your curriculum and your goals.
Are you interested in a free hands-on workshop to walk you through the process step by step? Reach out to us.
And if you’re ready to explore more AI tools and practical applications for higher education, talk to your institution about Generative AI in Practice: Essential Skills for Higher Ed Professionals (group offering). We’ll cover tools, strategy, guardrails, and real implementation ideas you can use right away.
