StudyRescue AI

An autonomous AI agent that detects when students fall behind and automatically rebuilds their study plan to get them back on track.

Comment

Inspiration

Students often reach exam time with multiple subjects, weak topics, and very limited time. The difficult part is not only learning the material — it is deciding what to study first, how much time to spend, and how to recover efficiently.

I built StudyRescue AI to help students turn an overwhelming exam situation into a practical, personalized recovery plan.

What it does

StudyRescue AI is an agentic study-recovery assistant that:

  • Understands the student's subjects, weak topics, available study time, and days remaining.
  • Prioritizes the most important weak areas.
  • Generates a personalized day-wise recovery plan.
  • Saves the generated plan to Google Cloud Firestore.
  • Retrieves the saved plan when the student asks for it later.

This makes the system more than a simple chatbot: it performs a workflow and maintains persistent study-plan data.

How I built it

The project was built using Google Agent Development Kit (ADK) with Gemini 3.6 Flash as the AI model.

The main workflow is:

Student → StudyRescue Agent → Gemini → Recovery Plan → Firestore → Saved Plan → Retrieval

Google Cloud Firestore is used to persist the student's study plan so that it can be retrieved in a later interaction.

Example

A student can provide:

I have 3 days before my exams. My subjects are DBMS, COA, and Data Structures. My weak topics are normalization, transactions, cache memory, and trees. I can study 4 hours per day.

StudyRescue generates a structured recovery plan, for example:

  • Day 1: DBMS — Normalization and Transactions
  • Day 2: COA — Cache Memory and Data Structures — Trees
  • Day 3: Comprehensive revision and practice

The plan is then saved to Firestore.

The student can later ask:

Show me my saved study plan.

The agent retrieves the previously stored plan.

What I learned

Building StudyRescue AI helped me understand how to build an agentic application rather than only a conversational AI application.

I learned about:

  • Google ADK agent development
  • Gemini model integration
  • Tool-based agent workflows
  • Google Cloud Firestore
  • Persistent application data
  • Environment variables and credential management
  • Structuring an AI application for a real-world use case

Challenges

One of the main challenges was connecting the agent workflow with Google Cloud services while keeping credentials secure.

I also had to troubleshoot local development issues involving the Python virtual environment, ADK Web, Google Cloud configuration, Firestore setup, and model availability.

These challenges helped me understand the practical side of deploying an AI agent beyond writing the initial code.

Why it matters

StudyRescue AI focuses on a simple but common student problem: falling behind when exam time is running out.

Instead of giving generic motivation or study tips, the agent converts the student's constraints into an actionable recovery plan and remembers that plan for later.

Future improvements

Future versions could include:

  • Progress tracking
  • Daily study reminders
  • Calendar integration
  • Adaptive plans based on completed topics
  • Exam timetable integration
  • Personalized learning-resource recommendations

Conclusion

StudyRescue AI demonstrates how an agent can combine reasoning, actions, and persistent cloud data to solve a practical student problem.

Study Smarter. Recover Faster. Achieve More.

Built With

Share this project:

Updates

Submission history