How to build an AI Course Assistant

TL;DR

What it does:
This agent answers student questions about a specific university course by searching across course documents, Google Drive, and SharePoint, then generates a concise, well-formatted answer. It can also create a task in Asana based on the AI’s response.

Who it’s for:

• University students in the course

• Course instructors or TAs managing student queries

• Academic support staff

Time to value:
Immediate—students get instant, context-rich answers to course questions, and staff can automate follow-up actions (like creating tasks).

Output:
A markdown-formatted answer to the student’s question, citing course materials and knowledge bases, and (optionally) a new task on a workload management platform with the answer as its title.

Common Pain Points For Onboarding Students

• Students struggle to find up-to-date, relevant information across multiple document sources (syllabus, schedules, exams, etc.)

• Answers to logistical or deadline questions are buried in different platforms (Google Drive, SharePoint, etc.)

• Manual triage and follow-up actions (like creating tasks) are time-consuming for staff

• Inconsistent or incomplete responses due to fragmented knowledge

What This Agent Delivers

• Unified search across course documents, Google Drive, and SharePoint

• AI-generated, concise, and friendly answers tailored to student questions

• Consistent formatting and citation of sources

• Automatic creation of follow-up tasks (e.g., for unresolved or important queries)

• Reduced manual workload for instructors and support staff

See the Agent in Action:

Step-by-Step Build (StackAI nodes)

1) Student Question

What it does:

• Receives the student’s question as text input

Goal:

• Capture the user’s query to drive the rest of the workflow

2) Course Documents

What it does:

• Searches a curated set of course files (syllabus, exams, schedule, etc.) for relevant information

Goal:

• Provide authoritative, course-specific context for the AI’s answer

3) Google Drive

What it does:

• Searches connected Google Drive for additional course-related content

Goal:

• Expand the context pool to include any relevant files stored in Google Drive

4) SharePoint 2

What it does:

• Searches connected SharePoint for course materials

Goal:

• Ensure no relevant information is missed, even if stored in SharePoint

5) Anthropic

What it does:

• Receives the student’s question and all retrieved context from the three knowledge bases

• Generates a brief, polite, and conversational answer, citing sources as needed

Goal:

• Synthesize a high-quality, student-friendly response using all available information

Instructions

You are an AI designed to assist students with inquiries related to the "AI Transformation & Business Principles" course, taught by Prof. Michaels in Fall 2025.



- You will receive context from a comprehensive knowledge base containing course materials, syllabi, reading lists, calendars, and more.

- Provide brief and polite responses to student questions regarding course logistics, deadlines, and project expectations.

- Maintain a conversational and friendly tone in your interactions

Prompt

Context to answer user questions

{knowledgebase-0}

{knowledgebase-2}

{knowledgebase-1}



User message

{in-0}

6) Template

What it does:

• Formats the AI’s answer and the original question in a clear, markdown-styled output

• Adds a note that the answer was generated using course documents and knowledge bases

Goal:

• Ensure the response is easy to read and consistently branded

7) Create Task

What it does:

• Creates a new task in Asana with the AI’s answer as the task name

• For example, to track follow-up or escalate complex questions

• Can be changed to suit your course's workload management platform

Goal:

• Automate administrative follow-up, reducing manual work for staff

8) Answer

What it does:

• Presents the formatted answer to the student/user

Goal:

• Deliver the final, polished response