This guide helps you get the most out of your AI chatbots. Following these practices will improve student experiences and make your course management easier.
Organizing Your Knowledge Base
Why Organization Matters
The quality of your chatbot responses depends directly on how well you organize your course materials. A well-organized knowledge base helps the AI find the right information quickly and give accurate answers. A messy knowledge base leads to confused or incomplete responses.
Think of it like a filing cabinet: if everything is thrown in one drawer, it takes longer to find what you need. But if documents are sorted into labeled folders, finding the right information is quick and easy.
Use Descriptive Titles for Your Documents
The title you give each document matters because students will see it in chatbot responses. When the chatbot answers a question, it often tells the student where the information came from by showing the document title.
Bad titles:
- document1.pdf
- notes.txt
- lecture.pdf
- scan001.pdf
Good titles:
- Chapter 3 - Thermodynamics Principles.pdf
- Week 5 Lecture - Database Normalization.txt
- Lab Safety Procedures - Updated Fall 2025.pdf
- Midterm Review - Key Concepts.pdf
When a student asks about thermodynamics and sees "Source: Chapter 3 - Thermodynamics Principles.pdf" they know exactly where to look for more information.
Create Separate Knowledge Bases by Topic
Instead of putting all your materials in one knowledge base, create separate knowledge bases for different topics or units.
Example organization for an Introduction to Biology course:
Knowledge Base: Unit 1 - Cell Structure
- Cell membrane overview.txt
- Organelles and their functions.pdf
- Cell division lecture notes.txt
Knowledge Base: Unit 2 - Genetics
- DNA and RNA basics.pdf
- Mendelian inheritance.txt
- Genetic disorders overview.pdf
Knowledge Base: Unit 3 - Evolution
- Natural selection principles.pdf
- Evidence for evolution.txt
- Speciation lecture notes.pdf
Knowledge Base: Lab Materials
- Lab safety procedures.pdf
- Lab report template.txt
- Equipment guide.pdf
This organization lets you create focused chatbots and makes it easier to update materials for specific topics.
Upload Transcripts and Outlines
Lecture recordings are valuable, but the AI cannot watch videos. To make your lecture content available to students through the chatbot:
- Upload lecture transcripts
- Many video platforms can auto-generate transcripts
- Edit transcripts to fix errors and add clarity
- Include timestamps if helpful for students
- Create lecture outlines
- Summarize main points from each lecture
- Include key terms and definitions
- Add examples discussed in class
- Convert slides to text
- Export your presentation notes
- Add context that you would normally say out loud
- Include explanations for diagrams and charts
Transcripts and outlines are often more useful than raw slides because they capture the explanations you give during class.
Creating an Errata Document
Instead of editing every document when you find an error or want to add a clarification, create an errata document with keyword triggers. This saves time and ensures corrections are always applied.
How to create an errata document:
- Create a new text file called "Course Corrections and Updates.txt"
- Write entries that include the keywords students might use
- Add this errata document to your knowledge base
- When students ask about these topics, the chatbot will include the corrections in its responses
- Update this single document whenever you need to make corrections, rather than editing multiple files
Example errata entries:
---
CORRECTION - Midterm Exam Date
The midterm exam date mentioned in the syllabus is incorrect.
The correct date is: November 15, 2025 (not November 8).
Keywords: midterm, exam, test date, November
---
CLARIFICATION - Problem Set 3, Question 7
There is a typo in Problem Set 3, Question 7. The equation should read:
F = ma (not F = Ma). Use lowercase m for mass.
Keywords: problem set 3, question 7, PS3, equation, force
---
UPDATE - Office Hours Change
Starting Week 8, office hours have moved to Wednesdays 2-4pm.
This replaces the Tuesday time slot listed in the syllabus.
Keywords: office hours, meeting time, help, Tuesday, Wednesday
---
Creating Effective Chatbots
Why Focused Chatbots Work Better
A common mistake is creating one chatbot that contains all your course materials. While this seems convenient, it actually makes the chatbot less effective.
Problems with one big chatbot:
- Responses are slower because there is more material to search
- The AI may pull information from unrelated topics
- Students get overwhelmed with too much information
- Harder to maintain and update
Benefits of focused chatbots:
- Faster, more accurate responses
- Students get relevant information without distractions
- Easier to update specific topics
- You can track which topics students need most help with
Recommended chatbot structure for a course:
General Course Chatbot
Linked to: Syllabus, course policies, FAQ
Purpose: Answer administrative questions
Unit 1 Study Helper
Linked to: Unit 1 knowledge base only
Purpose: Help with Unit 1 concepts
Unit 2 Study Helper
Linked to: Unit 2 knowledge base only
Purpose: Help with Unit 2 concepts
Lab Assistant
Linked to: Lab materials knowledge base
Purpose: Help with lab procedures and reports
Exam Review Bot
Linked to: Review materials, practice problems
Purpose: Help students prepare for exams
Writing Good Chatbot Descriptions
The description you write for each chatbot helps students choose the right one. Be specific about what the chatbot can help with.
Bad description:
"Ask me anything about the course."
Good description:
"I can help you understand concepts from Unit 2: Genetics. Ask me about DNA structure, RNA transcription, Mendelian inheritance, or genetic disorders. For questions about other units, please use a different chatbot."
This tells students exactly what to expect and reduces frustration when the chatbot cannot answer questions outside its scope.
When to Publish and Unpublish Chatbots
Use the publish and unpublish features strategically:
- Keep chatbots unpublished while you are still setting them up and testing
- Publish chatbots at the start of the relevant unit or week
- Unpublish chatbots for past units if you want students to focus on current material
- Unpublish review chatbots after exams are complete
This keeps the student interface clean and focused on what matters right now.
Using Guided Chatbots
The Power of Guided Conversations
Guided chatbots are different from regular chatbots. Instead of free-form question and answer, they walk students through a structured experience that you design step by step.
Think of guided chatbots as interactive tutorials where you control the flow of information and can check understanding along the way.
Use Cases for Guided Chatbots
1. Interactive Tutorials
Walk students through complex procedures step by step.
Example: "Setting Up Your Development Environment"
- Step 1: Welcome and overview of what we will install
- Step 2: Download instructions for the required software
- Step 3: Installation walkthrough with common troubleshooting tips
- Step 4: Verification that installation worked correctly
- Step 5: First project setup
2. Concept Walkthroughs
Explain difficult concepts in a structured way with built-in comprehension checks.
Example: "Understanding Photosynthesis"
- Step 1: Introduction - What is photosynthesis?
- Step 2: The light reactions explained
- Step 3: Check your understanding (ask student to explain in their words)
- Step 4: The Calvin cycle explained
- Step 5: Check your understanding
- Step 6: How the two processes connect
- Step 7: Summary and key takeaways
3. Textbook Replacement
Convert dense textbook chapters into interactive conversations. This is more engaging than reading because students participate actively.
Example: "Chapter 4 - The Constitution"
- Step 1: Why was a new constitution needed?
- Step 2: The key debates at the Constitutional Convention
- Step 3: What do you think were the main concerns? (student responds)
- Step 4: The Great Compromise explained
- Step 5: The Three-Fifths Compromise and its implications
- Step 6: Ratification process
- Step 7: Review questions
4. Lab Preparation
Ensure students are ready before they enter the lab.
Example: "Pre-Lab Checklist - Titration Experiment"
- Step 1: Safety requirements review
- Step 2: Equipment you will need
- Step 3: Procedure overview
- Step 4: What is the purpose of this experiment? (student responds)
- Step 5: Common mistakes to avoid
- Step 6: Confirmation that student is ready
5. Assignment Walkthroughs
Guide students through complex assignments.
Example: "Research Paper Guidelines"
- Step 1: Assignment overview and requirements
- Step 2: Choosing a topic - criteria and examples
- Step 3: Research strategies and source requirements
- Step 4: Outline structure expectations
- Step 5: Citation format review
- Step 6: Common mistakes to avoid
- Step 7: Submission instructions
Designing Effective Guided Chatbot Steps
Keep each step focused on one idea:
- One concept per step
- Clear, concise language
- Include examples when helpful
Include interaction points:
- Ask students to explain concepts back
- Have them predict what comes next
- Let them ask clarifying questions
Build in comprehension checks:
- Add steps that ask students to summarize
- Include practice questions
- Let students indicate if they need more explanation
End with clear next steps:
- Summarize key points
- Tell students what to do next
- Point them to related resources
Quiz Helper Chatbots
Setting Up Quiz Preparation Chatbots
Quiz helper chatbots are some of the most popular with students. Here is how to set them up effectively:
- Create a dedicated knowledge base
Include only materials relevant to the quiz:
- Study guide
- Practice problems with solutions
- Key concepts summary
- Common mistakes to avoid
- Sample quiz questions (from previous terms if appropriate)
- Write a clear description
Example: "I help you prepare for Quiz 2, which covers Chapters 4-6. I can explain concepts, work through practice problems, and help you identify areas where you need more study time."
- Include worked examples
Upload documents that show step-by-step problem solving:
- Show the problem
- Show each step of the solution
- Explain why each step is taken
- Highlight common errors
- Add a tips document
Create a document with quiz-taking strategies:
- Time management suggestions
- How to approach different question types
- What to do if you get stuck
- Key formulas or concepts to memorize
Timing Your Quiz Chatbots
- Publish the quiz helper chatbot about one week before the quiz
- This gives students time to use it for studying
- Unpublish after the quiz to keep focus on current material
- You can republish before final exams for comprehensive review
Anonymous Chatbots for Public Access
When to Use Anonymous Chatbots
Anonymous chatbots let anyone chat without logging in. They are ideal for:
- Course Website FAQs - Embed a chatbot that answers common questions about your course
- Prospective Students - Share a link so people considering your course can ask questions
- Open Office Hours - Provide 24/7 AI assistance without requiring enrollment
- LMS Integration - Embed chat in Canvas, Moodle, or other systems where students are already authenticated
- Event Support - Share a temporary chatbot for conferences or workshops
Setting Up Anonymous Chatbots Effectively
1. Start with a Good Base Chatbot
Anonymous chatbots inherit settings from a base chatbot. Set up the base chatbot first:
- Choose content appropriate for public visitors (avoid internal course details)
- Configure knowledge bases with publicly relevant information
- Test the base chatbot thoroughly before making it public
2. Write a Clear Public Description
The description is what visitors see first. Make it helpful:
Bad description: "Ask me anything"
Good description: "Ask me about course topics, prerequisites, or enrollment - I am here to help you decide if this course is right for you"
3. Add Custom Instructions for the Public Audience
Anonymous chatbots can have their own instructions on top of the base chatbot:
- Tell the AI that users are not enrolled students
- Instruct it to avoid sharing answers to assignments or exams
- Guide it to provide general course information rather than specific grades or deadlines
Example custom instructions:
"You are talking to prospective students or the general public, not enrolled students. Do not provide answers to assignments, quizzes, or exams. Focus on explaining course concepts, prerequisites, and what students can expect to learn. If asked about enrollment deadlines or registration, direct them to the registrar's website."
4. Control Access with Publish/Unpublish
- Keep anonymous chatbots unpublished while testing
- Publish only when you are ready for public access
- Unpublish if you need to make changes or take a break
- Remember: unpublishing stops all public access immediately
Embedding Chatbots on Your Website
Anonymous chatbots can be embedded as iframes on any website:
- Copy the embed code from the anonymous chatbot settings
- Paste it into your course website, syllabus page, or LMS
- Adjust the width and height as needed for your layout
Common embedding locations:
- Course landing page - help visitors learn about the course
- FAQ page - let the AI answer common questions
- Office hours page - provide 24/7 AI support alongside scheduled hours
- Syllabus - embed at the bottom for quick reference
Monitoring Anonymous Chatbot Usage
Keep an eye on how your anonymous chatbots are being used:
- Session counts - See how many people are using the chatbot
- Message counts - Understand engagement depth
- 7-day graphs - Spot usage trends and peaks
- Chat history - Review conversations to improve the experience
If usage is low:
- Make sure the link or embed is visible
- Promote the chatbot in course announcements or on your website
- Check that the description is clear about what the chatbot can help with
AIAS and Tutor Assignment Chatbots
When to Use AIAS vs Standard Chatbots
Use standard chatbots when students need open-ended help with course material. Use AIAS chatbots when you want structured AI use with documented process evidence and grading. Choose tutor assignments specifically when you want students to demonstrate understanding by teaching the AI rather than being taught by it.
Choosing the Right Engagement Type
Each of the 6 tutor assignment engagement types targets a different pedagogical skill:
- Explain to Learn (Feynman) — Best for checking conceptual understanding. Students must articulate ideas clearly enough to teach them.
- Argue Your Case — Best for developing argumentation skills. Students must defend their positions with evidence and reasoning.
- Find My Mistakes — Best for developing critical analysis. Students must identify errors and explain corrections.
- Guide My Solution — Best for problem-solving courses. Students must break down problems and guide the AI step by step.
- Simplify for Me — Best for testing depth of understanding. Students must explain complex concepts in accessible language.
- Misconception Check — Best for identifying knowledge gaps. Students must distinguish correct from incorrect statements and explain why.
Preparing Knowledge Bases for Tutor Assignments
- Include comprehensive course materials so the AI can play its role convincingly
- For "Find My Mistakes" engagements, ensure the knowledge base covers common misconceptions
- For "Argue Your Case" engagements, include materials with multiple perspectives or debatable points
- Test the chatbot yourself to verify the AI's learner persona is appropriate for your content
Tips for Effective Use
- Explain the purpose of the tutor assignment to students before they begin — learning by teaching is most effective when students understand why
- Start with "Explain to Learn" or "Simplify for Me" for introductory courses, and use "Argue Your Case" or "Find My Mistakes" for advanced courses
- Review the engagement-specific reflection questions on the evidence form to ensure they align with your learning objectives
- Use the grading dashboard to compare how students perform across different engagement types
- Consider assigning different engagement types for different topics to develop a range of skills
Driving Student Engagement
Getting Students to Use the Chatbots
Students will not automatically use the chatbots just because they exist. You need to actively promote them.
- Introduce chatbots on the first day
- Demonstrate how to access them
- Show a sample conversation
- Explain what they can and cannot do
- Integrate into assignments
- "Use the Unit 2 chatbot to check your understanding before submitting"
- "The Lab Assistant chatbot can help you troubleshoot common issues"
- "Review the exam prep chatbot before your study session"
- Mention in lectures
- "If you are confused about this topic, the chatbot has great examples"
- "I have added the slides to the chatbot, so you can ask questions later"
- "The guided tutorial will walk you through this process step by step"
- Reference in office hours
- "Have you tried asking the chatbot? Let me show you how"
- "The chatbot has practice problems that might help"
- "For questions like this, the chatbot is available 24/7"
- Send reminders
- Before exams: "Remember to use the exam review chatbot"
- When new content is added: "I have updated the Unit 3 chatbot with today's lecture"
- When students struggle: "Many students find the guided tutorial helpful"
Monitoring and Improving Usage
Check the Engagement tab regularly to see:
- Which chatbots are being used most
- When students are most active
- Whether usage increases before exams
If a chatbot is not being used:
- Make sure students know it exists
- Check if the description is clear
- Verify the knowledge base has useful content
- Consider if students need that chatbot at all
Using Insights Effectively
What Insights Can Tell You
The Insights tool analyzes student conversations to identify:
- What topics students struggle with most
- Common misconceptions
- Questions that come up repeatedly
- Patterns in how students learn
This information helps you improve your teaching, not just the chatbots.
When to Generate Insights
Generate insights at strategic times:
Weekly Check-ins
- Generate "New Activity" insights each week
- Look for emerging confusion before it becomes widespread
- Identify topics to review in the next lecture
Before Exams
- Generate insights for the exam period
- See what students are studying
- Identify topics that need last-minute review sessions
End of Unit
- Generate "Full History" insights for each unit
- Understand overall comprehension
- Identify improvements for next term
End of Term
- Generate comprehensive insights
- Evaluate overall chatbot effectiveness
- Plan improvements for next term
Using Custom Filters for Deeper Analysis
Custom filters let you focus on specific aspects:
By Time Period
- Compare early semester vs late semester engagement
- See how study patterns change before exams
- Track improvement over time
By Chatbot
- Which chatbot gets the most questions?
- Which chatbot has the most confused students?
- Are students using the right chatbot for their questions?
By Student
- Identify students who may need extra help
- See which students are highly engaged
- Track individual progress over time
Acting on Insights
Insights are only valuable if you act on them:
- If students are confused about a topic:
- Add more materials to the knowledge base
- Create a guided tutorial
- Address the confusion in your next lecture
- If a question comes up repeatedly:
- Add a FAQ document to the knowledge base
- Create an errata entry if there is a misunderstanding
- Consider if your original materials are clear
- If students are asking questions the chatbot cannot answer:
- Add the missing content to the knowledge base
- Check if students are using the wrong chatbot
- Consider creating a new chatbot for that topic
Monitoring Student Prompts
Why Check Student Prompts
The Student Prompts tab shows you exactly what students are asking. This is valuable for understanding:
- What students are confused about
- How students phrase their questions
- What topics generate the most questions
- Whether students are using chatbots appropriately
How to Review Student Prompts Effectively
Regular Quick Checks
- Spend 5-10 minutes a few times per week
- Scan recent prompts for concerning patterns
- Note topics that come up repeatedly
Before Lectures
- Check prompts related to upcoming topics
- See what prerequisite concepts may need review
- Identify questions to address proactively
Before Exams
- Look at what students are asking about
- Identify widespread confusion to address in review sessions
- See if students are using exam prep resources effectively
Using Filters to Find What Matters
Filter by Date
- Focus on recent activity
- Compare different time periods
- Track changes over the semester
Filter by Chatbot
- See which chatbot gets which types of questions
- Identify if students are using the wrong chatbot
- Understand how each chatbot is being used
Filter by Student
- Check on students you are concerned about
- See how engaged different students are
- Identify students who might need extra support
Use Search
- Search for specific topics or keywords
- Find all questions about a particular concept
- Track how a specific confusion evolves over time
What to Do With What You Learn
- Common Questions
If you see the same question multiple times:
- Add clearer content to the knowledge base
- Address it in class
- Create an FAQ entry
- Student Struggles
If a student seems confused across multiple prompts:
- Reach out to offer help
- Suggest specific resources
- Consider if they need additional support
- Chatbot Limitations
If students are asking questions the chatbot cannot handle:
- Add missing content
- Adjust chatbot descriptions to set expectations
- Create a new chatbot if needed
- Course Improvements
Use patterns in student prompts to:
- Identify unclear parts of your materials
- Find topics that need more class time
- Improve explanations for next term
Summary of Best Practices
Knowledge Base
- Use descriptive document titles
- Organize materials into focused knowledge bases
- Upload transcripts and outlines of lectures
- Create an errata document for corrections
Chatbots
- Create focused chatbots, not one big one
- Write clear, specific descriptions
- Publish and unpublish strategically
- Test before making available to students
Guided Chatbots
- Use for tutorials, walkthroughs, and interactive learning
- Keep each step focused on one idea
- Include comprehension checks
- Consider as a textbook alternative for complex topics
Quiz Helpers
- Create dedicated knowledge bases for each quiz
- Include worked examples and practice problems
- Publish about a week before the quiz
Anonymous Chatbots
- Use for public FAQs, prospective students, and website embeds
- Start with a well-tested base chatbot
- Add custom instructions for public audience
- Write clear descriptions of what the chatbot can help with
- Monitor usage and review chat history periodically
AIAS and Tutor Assignments
- Use AIAS for structured AI use with process evidence; use tutor assignments when students should learn by teaching
- Match engagement types to pedagogical goals and course level
- Prepare knowledge bases with comprehensive materials for the AI's learner persona
- Explain the purpose to students and review engagement-specific reflections
Driving Engagement
- Introduce chatbots on day one
- Integrate into assignments and lectures
- Send reminders at key times
- Monitor usage and adjust
Insights
- Generate weekly and at key milestones
- Use custom filters for deeper analysis
- Act on what you learn
Student Prompts
- Check regularly for patterns
- Use filters to focus your review
- Respond to what you learn
Following these practices will help you create an effective learning environment where students can get help whenever they need it.