Vacation Idea Generator - AI Travel Assistant

A Go-based microservice that generates personalized vacation ideas using AI (Ollama). This project demonstrates how to integrate LLMs into a web service using Gin framework and LangChain Go.

Features

• AI-Powered Vacation Planning: Generate personalized travel itineraries based on user preferences
• Async Processing: Non-blocking vacation idea generation
• RESTful API: Clean HTTP endpoints for creating and retrieving vacation ideas
• Ollama Integration: Local LLM processing with configurable models
• Real-time Status: Check generation progress and completion status

Stack

• Backend: Go 1.21+
• Web Framework: Gin
• AI/ML: LangChain Go + Ollama
• Configuration: Environment variables + .env file
• UUID Generation: Google UUID

Prerequisites

Before running this project, ensure you have:

1. Go 1.21+ installed
2. Ollama installed and running
3. At least one LLM model pulled in Ollama (e.g., llama3:latest)

Installing Ollama

# On macOS
brew install ollama

# On Linux
curl -fsSL https://ollama.ai/install.sh | sh

# On Windows
# Download from https://ollama.ai/download

Setting up Ollama Models

# Start Ollama service
ollama serve

# In a new terminal, pull a model
ollama pull llama3:latest

# Verify the model is available
ollama list

Installation

1. Clone the repository

  git clone https://github.com/bianavic/go-langchain.git
  cd go-langchain

2. Set up environment variables

Create a .env file in the project root:

# ollama base URL
LLM_BASE_URL=http://localhost:11434

# The LLM you want to use for the agents
LLM_MODEL=llama3:latest

3. Initialize module and install dependencies

go mod init github.com/bianavic/go-langchain
go mod init

4. Run the application

  cd cmd
  go run main.go

API Usage

Important Notes About Data

🔔 This is a demo service - Data is stored in memory only:

We're using in-memory storage as a deliberate simplification to focus on the core AI integration without database complexity

• Vacation ideas disappear when the server restarts
• Perfect for testing and development
• Not suitable for production use without adding a real database

1. Create a Vacation Idea

• Endpoint: POST /vacation/create

  curl --location 'localhost:8080/vacation/create' \
  --header 'Content-Type: application/json' \
  --data '{
  	  "favorite_season": "winter",
      "hobbies": ["visiting museums", "ice skating"],
      "budget": 1000
  }'

# Response: {"id":"bdd32ff6-7e56-4c36-8a63-ef1a0ab7fcbf","completed":false}

2. Check Vacation

• Endpoint: GET /vacation/:id

This endpoint waits (up to 5 minutes) for the vacation idea to be fully generated before returning the response.

  curl --location 'localhost:8080/vacation/bdd32ff6-7e56-4c36-8a63-ef1a0ab7fcbf'

## Response: 
#{
#    "id": "bdd32ff6-7e56-4c36-8a63-ef1a0ab7fcbf",
#    "completed": true,
#    "idea": "ollama response goes here...."
#}