💡[Feature]: Day 3: AI ChatBot Bakend: Implement ChatBot Backend (Generative AI Integration)

[SB2318]

Is there an existing issue for this?

• I have searched the existing issues

Feature Description

Requirement: Please complete the task before #629

🧠 Why Chatbot Backend?

To communicate with Google Gemini (Generative AI Model) using your Gemini API Key, we need to build a backend layer that works as a mediator service.

This backend becomes your Application Programming Interface (API) responsible for:

• Sending user prompts to the Gemini Generative AI Model
• Receiving model-generated responses
• Managing message storage and conversation history
• Acting like a real-time LLM-powered messaging system

Instead of chatting with another human, your server will continuously send queries to Gemini's text-generation endpoint and return AI-generated replies.

I’ve prepared some initial setup along with basic endpoints. Feel free to use them as a starting point while building the rest. (Below is the link).

https://github.com/SB2318/IEEE-s-Mindful-Devs-Bootcamp/tree/main/chatbot

Visualization

https://uhsocial.in/docs/#/ChatBot/post_gemini_send

Use Case

✅ Tasks

• Fork the submodule (https://github.com/SB2318/IEEE-s-Mindful-Devs-Bootcamp)
• Create a new branch add your changes there
• Create PR For the submodule

---

✅ Endpoints to Implement (Phase 1)

1️⃣ POST /send-message

Purpose:
Send a user prompt to the Gemini Generative AI model and return the generated response.

Required Input

• userId
• conversationId
• text – user prompt / question

Workflow

1. Backend receives the user's message
2. Calls the Gemini Generative AI API using the model endpoint
3. Saves both:
   • User’s prompt
   • Gemini’s generated response
4. Returns the AI-generated output to frontend

Notes

This behaves as a normal messaging system, except the reply is generated by the LLM (Gemini).

---

2️⃣ GET /load-conversations

Purpose:
Fetch all messages for a specific user with proper ordering.

Required Input

• userId

Workflow

1. Query all conversations from database (based on userId)
2. Sort messages by timestamp
3. Return complete conversation history

Notes

This endpoint helps in:

• Displaying conversation history
• Resuming past LLM sessions
• Maintaining context for the Generative AI model

---

📝 Future Endpoints (Optional but Useful)

• POST /regenerate-message (ask model again)
• POST /upload-file (for Gemini document/vision capabilities)

---

✔️ Acceptance Criteria

• Gemini API integrated using official SDK or REST
• Messages stored in DB (MongoDB recommended)
• Conversation history fetched properly
• Clean architecture (Controller → Service → Routes)
• Proper error handling for model failures
• API key securely stored in environment variables

---

Benefits

Learning

Add ScreenShots

Priority

High

Record

• I have read the Contributing Guidelines
• I'm a GSSOC'24 contributor
• I'm a IEEE IGDTUW contributor
• I want to work on this issue

[github-actions]

Thank you for creating this issue! We'll look into it as soon as possible. Your contributions are highly appreciated! 😊