SETUP_GUIDE.md

Agent Bot - Setup and Quick Start Guide

Current Status (Fixed)

✅ All import errors have been resolved ✅ Dependencies installed correctly ✅ Code adapted for LangChain 0.1.6 compatibility

What Was Fixed

1. LangChain Version Mismatch

• Problem: Code was using create_tool_calling_agent from newer LangChain (1.0+), but virtualenv had incompatible version
• Solution:
  • Downgraded to langchain==0.1.6 which has AgentExecutor and initialize_agent
  • Updated services/agent_core.py to use initialize_agent() with AgentType.CHAT_ZERO_SHOT_REACT_DESCRIPTION

2. Missing Dependencies

• Problem: Missing sentence-transformers, pinecone-client, torch, and other ML libraries
• Solution:
  • Created requirements.txt with all dependencies pinned
  • Installed complete dependency tree (~3GB+ with PyTorch)

3. Missing Singleton Function

• Problem: get_document_service() was not defined in services/document.py
• Solution: Added singleton pattern getter function at end of file

4. Missing Environment Configuration

• Problem: No .env file with API keys
• Solution: Created .env.example template

Next Steps to Start the Application

Step 1: Set Up Environment Variables

Create a .env file in the Agent_Bot directory:

cd /home/randitha/Desktop/IT/UoM/TechTorque-2025/Agent_Bot
cp .env.example .env

Then edit .env and add your actual API keys:

# Required keys:
GOOGLE_API_KEY=your_actual_google_gemini_api_key
PINECONE_API_KEY=your_actual_pinecone_api_key

Step 2: Start the Application

# Activate virtualenv (if not already active)
source .venv/bin/activate

# Or directly run with virtualenv python:
/home/randitha/Desktop/IT/UoM/TechTorque-2025/Agent_Bot/.venv/bin/python main.py

Step 3: Access the API

Once running, the service will be available at:

• Base URL: http://localhost:8091
• API Endpoint: http://localhost:8091/api/v1/ai/chat
• Health Check: http://localhost:8091/health
• API Docs: http://localhost:8091/docs

Where to Get API Keys

Google Gemini API Key

1. Go to: https://makersuite.google.com/app/apikey
2. Create a new API key for Gemini
3. Copy the key to your .env file

Pinecone API Key

1. Sign up at: https://www.pinecone.io/
2. Create a free "Starter" project
3. Go to "API Keys" in dashboard
4. Create/copy your API key
5. Create an index named techtorque-kb with dimension 384

Files Modified

• services/agent_core.py - Updated agent initialization for LangChain 0.1.6
• services/document.py - Added missing singleton getter function
• requirements.txt - Created with all dependencies
• .env.example - Created configuration template

Commit Your Changes

cd /home/randitha/Desktop/IT/UoM/TechTorque-2025/Agent_Bot
git add services/agent_core.py services/document.py requirements.txt .env.example
git commit -m "fix: Resolve LangChain import errors and add dependencies

- Adapt agent_core.py for LangChain 0.1.6 API (use initialize_agent)
- Add missing get_document_service() singleton function
- Create requirements.txt with pinned dependencies
- Add .env.example configuration template"

Testing Without API Keys (Optional)

If you don't have API keys yet but want to test imports, you can temporarily set dummy values:

export GOOGLE_API_KEY=dummy_key_for_testing
export PINECONE_API_KEY=dummy_key_for_testing
python main.py

The app will start but fail when actually trying to use the APIs. This is useful for verifying all imports work.

Architecture Notes

This Agent Bot is part of a microservices architecture:

• Port: 8091 (Agent Bot service)
• Dependencies: Authentication, Vehicle, Project, Time Logging, Appointment services
• Features:
  • LangChain-based AI agent with tool calling
  • RAG (Retrieval Augmented Generation) with Pinecone vector store
  • Integration with TechTorque backend microservices
  • Google Gemini 2.5 Flash model

Troubleshooting

Import Errors

✅ Fixed - LangChain version now matches code expectations

"Module not found" for sentence_transformers

✅ Fixed - All ML dependencies now installed

"GOOGLE_API_KEY not found"

⚠️ Action Required: Create .env file with actual API keys

Large Download Size

ℹ️ PyTorch and ML models are large (~3GB). This is normal for AI applications.