FnB-DocBot

A smart food and beverages bot that uses SOPs/policies to give response on user query and gives managers to the get dashboard access for completion, competency, and adherence.

Problem Statement

Design and prototype an internal-facing digital training solution—centered on a Document Intelligence Bot (docbot)—that:

• Converts existing SOPs/policies into just-in-time, stepwise guidance (e.g., “Deep fryer safety: step 1… step 2…”).
• Delivers microlearning (short lessons + quizzes), supports spaced repetition, and tracks progress & compliance.
• Gives managers dashboards for completion, competency, and adherence (audit-ready).
• Works on low-friction devices used in F&B (shared tablets, kiosk mode, mobile BYOD).

🚀 Features

• Docbot Chat that only answers from provided SOPs/policies (no hallucinations; must show citations).
• Procedure Mode (step-by-step, with checkboxes and “show me how” snippets/images if available).
• Quiz Engine (MCQ/true-false) + spaced repetition (e.g., Leitner).
• Manager Dashboard with: completion %, average score, last trained date, checklist adherence.
• User roles (Crew vs Manager) and basic auth (okta/email+otp/mock SSO acceptable).
• Content versioning: responses display “SOP vX.Y, date”.

🛠️ Tech Stack

• Frontend: ReactJs
• Backend: Appwrite and Python with FastAPI
• Vector Search: faiss
• Authentication: Appwrite Auth
• AI Integration: Groq AI
• Development: uvicorn for hot reloading

📋 Prerequisites

• Python
• Groq API Key

⚙️ Installation

1. Clone the repository

   git clone <repository-url>
   cd backend

2. Install dependencies

   python3 -m venv ./venv

   source venv/Scripts/activate

   pip install --no-cache-dir -r ./requirements.txt

3. Environment Setup Create a .env file in the root directory with the following variables:

   # AI (Groq)
   GROK_API_KEY=your_groq_api_key

🚀 Running the Application

1. Start the main server

   uvicorn run main:app --reload

2. Start the frotend application

   npm run dev

📝 API Endpoints

Response

• POST /api/inngest - upload a new file
• GET /api/query - query to get response

🔄 Response Processing Flow

1. Response Creation

   • User submits a file and a group of file
   • System creates a temp file

2. RAG Processing

   • Inngest triggers and creates vector embeddings based on that temp file
   • Then Query triggers it also creates vector embeddings
   • And then perform vector search or similarity search on them

3. AI Response

   • And based on that search results AI responds to that query

🔍 Troubleshooting

Common Issues

1. Port Conflicts If you see "address already in use" error:

   # Find process using port: 8000
   lsof -i :3000
   # Kill the process
   kill -9 <PID>

2. AI Processing Errors

   • Verify GROK_API_KEY in .env
   • Check API quota and limits
   • Validate request format

🙏 Acknowledgments

• Groq AI for AI processing
• Appwrite for authentication

Authors

• Khush Soni: Backend and AI Integration

  

• Priyanshu Panwar: Frontend and Appwrite Implemenation

  

• Nehal Jain: Code and Repository Maintainence