	In Browser
	StumbleUpon
	del.icio.us
	Google
	Google Buzz
	reddit
	LinkedIn

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Student Tutoring Agent

Student Tutoring Agent

Author: Venkata Sudhakar

India has over 250 million school and college students, yet access to quality tutoring is limited by geography and cost. A Student Tutoring Agent built with ADK and Gemini can act as a personal tutor - explaining concepts at the right level, asking follow-up questions to test understanding, tracking which topics the student struggles with, and adapting the next lesson accordingly. This is the core of personalised learning at scale.

This tutorial builds a Student Tutoring Agent using Google ADK and Gemini 2.0 Flash. The agent teaches a concept, quizzes the student, evaluates the answer, and recommends the next topic based on performance.

The below example shows the Student Tutoring Agent teaching a Class 10 student the concept of compound interest for board exam preparation.

from google.adk.agents import Agent
from google.adk.tools import FunctionTool
from google.adk.runners import Runner
from google.adk.sessions import InMemorySessionService
from google.genai.types import Content, Part

# Topic curriculum map
CURRICULUM = {
    "compound_interest": {
        "prerequisites": ["simple_interest", "percentages"],
        "next_topics": ["loans_and_emi", "depreciation"],
        "exam_weightage": "5-8 marks (CBSE Class 10)",
        "formula": "A = P(1 + r/n)^(nt), CI = A - P"
    },
    "simple_interest": {
        "prerequisites": ["basic_arithmetic"],
        "next_topics": ["compound_interest"],
        "exam_weightage": "3-5 marks",
        "formula": "SI = (P x R x T) / 100"
    }
}

def get_topic_info(topic: str, student_class: int) -> dict:
    """Fetch curriculum info for a topic."""
    info = CURRICULUM.get(topic.lower().replace(" ", "_"), {})
    if not info:
        return {"error": "Topic not found in curriculum: " + topic}
    return {
        "topic": topic,
        "student_class": student_class,
        "formula": info.get("formula", ""),
        "exam_weightage": info.get("exam_weightage", ""),
        "prerequisites": info.get("prerequisites", []),
        "next_topics": info.get("next_topics", [])
    }

def evaluate_answer(topic: str, question: str,
                    student_answer: str, correct_answer: str) -> dict:
    """Evaluate a student answer and provide feedback."""
    # Simple keyword match for demonstration
    student_lower = student_answer.lower()
    correct_lower = correct_answer.lower()
    # Extract key numbers/concepts
    score = 0
    if any(word in student_lower for word in correct_lower.split()[:3]):
        score += 50
    if len(student_answer) > 20:  # student gave a detailed answer
        score += 25
    if any(c.isdigit() for c in student_answer):  # included calculation
        score += 25
    performance = "excellent" if score >= 90 else ("good" if score >= 60 else "needs_practice")
    return {
        "topic": topic,
        "score": score,
        "performance": performance,
        "correct_answer": correct_answer,
        "recommendation": "Move to next topic" if score >= 70 else "Revise " + topic + " with more examples"
    }

topic_tool = FunctionTool(func=get_topic_info)
eval_tool = FunctionTool(func=evaluate_answer)

tutor_agent = Agent(
    model="gemini-2.0-flash",
    name="student_tutoring_agent",
    instruction="""
        You are a friendly Student Tutoring Agent for Indian school students.
        When a student wants to learn a topic:
        1. Fetch topic info including formula and exam weightage
        2. Explain the concept clearly with a simple Indian context example
           (use rupees, Indian names, relatable scenarios)
        3. Ask one practice question
        4. When the student answers, evaluate it and give encouraging feedback
        5. Recommend the next topic based on performance
        Use simple language. Be encouraging. Relate to CBSE/ICSE exam patterns.
        Always show the formula and a step-by-step solved example.
    """,
    tools=[topic_tool, eval_tool]
)

session_service = InMemorySessionService()
runner = Runner(agent=tutor_agent, app_name="tutoring", session_service=session_service)

session = session_service.create_session(app_name="tutoring", user_id="student_raj")
message = Content(role="user", parts=[Part(text="I am in Class 10 and want to learn Compound Interest for my board exams.")])
for event in runner.run(user_id="student_raj", session_id=session.id, new_message=message):
    if event.is_final_response():
        print(event.content.parts[0].text)

It gives the following output,

Hi! Great choice - Compound Interest is worth 5-8 marks in CBSE Class 10!

COMPOUND INTEREST - The Formula
  A = P(1 + r/n)^(nt)    and    CI = A - P
  Where: P = Principal, r = annual rate (decimal),
         n = times compounded per year, t = time in years

Simple Example:
  Ravi deposits Rs 10,000 in SBI at 10% per year compounded annually for 2 years.
  Step 1: A = 10000 x (1 + 0.10/1)^(1x2)
  Step 2: A = 10000 x (1.10)^2
  Step 3: A = 10000 x 1.21 = Rs 12,100
  Step 4: CI = 12,100 - 10,000 = Rs 2,100
  Compare: Simple Interest would be only Rs 2,000!
  The extra Rs 100 is "interest on interest" - that is compound interest.

Now your turn! Solve this:
  Meena invests Rs 5,000 at 8% per annum compounded annually for 3 years.
  What is the Compound Interest earned?
  (Hint: Calculate A first, then CI = A - P)

The Student Tutoring Agent can be extended into a full adaptive learning system. Use ADK sessions to persist the student performance history across multiple topics and build a personalised revision schedule before exams. Integrate Gemini multimodal to accept handwritten solutions as images and provide feedback on the working. Deploy as a mobile app or WhatsApp bot for students in tier-2 and tier-3 cities.

Send your comments, suggestions or queries regarding this site to [email protected].