✈️ Intelligent Travel Planning Assistant - HelloAgents AMD ROCm Edition

Intelligent Travel Planning Assistant is an AI Agent application developed based on the HelloAgents framework that can run large language models locally on AMD 395 AI PCs. It combines the MCP protocol to call Amap APIs and automatically generates complete travel planning solutions. All computations are completed locally, protecting your privacy data.

HelloAgents Project: Link

In the following sections, I will guide you step by step to implement the setup and usage of the Intelligent Travel Planning Assistant, allowing AI to help you plan the perfect trip!

Project Highlights

🏠 Local Deployment: Run 30B parameter large language models on AMD 395 AI PC without internet connection
🔒 Privacy & Security: Sensitive data such as travel preferences and budget information are not uploaded to the cloud
🗺️ Real Data: Integrates Amap APIs to obtain real-time attraction and weather information
📝 Auto-generation: Outputs complete Markdown format travel planning documents
🤖 Agent Architecture: Based on HelloAgents framework, supports tool calling and multi-turn conversations

Technology Stack

AI Framework: HelloAgents - Simplifies Agent application development
Protocol Standard: MCP (Model Context Protocol) - Standard protocol for AI to call external services
Inference Platform: AMD ROCm - Open-source GPU computing platform
Model Deployment: Ollama / LM Studio / Linglong "Smart Linglong Classmate"
External Services: Amap API (attraction search, weather query)

Step 1: Environment Preparation

The basic environment for this guide is as follows:

----------------
AMD Ryzen™ AI 9 HX 395 Processor
Python 3.10+
Ollama / LM Studio / Linglong "Smart Linglong Classmate" (choose one)
----------------

This guide assumes learners are using an AMD 395 AI PC or other devices with AMD ROCm support. For information about local model deployment, please refer to Getting Started with ROCm Deploy

1.1 Install Python Dependencies

First, switch pip source for faster downloads and install dependency packages:

shell

# Upgrade pip
python -m pip install --upgrade pip

# Switch to faster PyPI source
pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple

# Install HelloAgents and related dependencies
pip install hello-agents requests python-dotenv uv

1.2 Obtain Amap API Key

Visit Amap Open Platform
Register and log in to your account
Enter the console and create an application
Obtain the API Key (select Web Service API)

Step 2: Local Model Deployment

Plan Selection

Choose one of the following three solutions based on your system and usage preferences:

Tool	Supported Platforms	Features	Recommended For
Ollama	Windows/Mac/Linux	CLI tool, lightweight, simple model mgmt	Developers familiar with CLI
LM Studio	Windows/Mac/Linux	GUI, visual operation, beginner-friendly	Users needing GUI
Linglong "Smart Linglong Classmate"	Linglong AI Workstation	Pre-installed, one-click deployment, optimized for AMD 395	Linglong Workstation Users

2.1 Using Ollama (Recommended)

Ollama is a lightweight local model management tool that supports AMD ROCm acceleration.

Install Ollama:

shell

# Visit the official website to download for Windows/Mac/Linux
# https://ollama.ai/

# Or install via command line (Linux)
curl -fsSL https://ollama.ai/install.sh | sh

Download and Run Models:

shell

# Download Qwen2.5 32B model (recommended)
ollama pull qwen2.5:32b

# Or download other models
ollama pull deepseek-r1:32b
ollama pull llama3.1:70b

Start Ollama Service:

shell

# Ollama will automatically run in the background
# Default port: http://localhost:11434
ollama serve

2.2 Using LM Studio

Download and install LM Studio
Search for and download models in the model library (Qwen2.5 32B recommended)
Start the service in the "Local Server" tab
Confirm the endpoint address (default: http://127.0.0.1:1234)
Select AMD ROCm as the inference engine in settings

2.3 Using Linglong "Smart Linglong Classmate"

If you are using a Linglong AI Workstation:

Open the pre-installed "Smart Linglong Classmate" application
Select an appropriate model (30B+ parameters recommended)
Start the local service with one click
Record the service endpoint address

Step 3: Configure the AMap API Key

This example launches the MCP service directly with uvx amap-mcp-server, so you do not need a separate mcp_config.json file or a standalone mcp_amap_server.py.

3.1 Install uv

shell

pip install uv

3.2 Configure the API Key

You can choose either approach:

Edit the amap_api_key value directly in travel_planner_mcp.py
Or set amap_api_key to empty in the script and inject the key through an environment variable

shell

setx AMAP_MAPS_API_KEY "your_amap_api_key_here"

Step 4: Run the Intelligent Travel Planning Assistant

4.1 Download Project Code

shell

# Clone the hello-rocm repository
git clone https://github.com/datawhalechina/hello-rocm.git

# Enter the project directory
cd hello-rocm/05-AMD-YES/05-hello-agents/smart-travel-planner

4.2 Configure Model Endpoint

Edit travel_planner_mcp.py and update the HelloAgentsLLM(...) configuration:

python

self.llm = HelloAgentsLLM(
    model="Qwen3-30B-2507-instruct",
    base_url="http://127.0.0.1:1234/v1",
    api_key="amd395"
)

If you are using a different local endpoint, just replace model and base_url with your own values.

4.3 Run the Assistant

shell

python travel_planner_mcp.py

4.4 Usage Example

The current sample script runs directly with a built-in Hangzhou 3-day example. To change the destination, budget, or preferences, edit the parameters in main():

python

result = planner.plan_travel(
    destination="杭州",
    days=3,
    budget=3000,
    preferences="自然风光和历史文化"
)

The AI assistant will automatically:

Call Amap APIs to search for attractions
Query weather information
Plan the itinerary based on budget and preferences
Generate a complete Markdown travel planning document

Step 5: View Generated Travel Plans

The generated travel plan is saved as a Markdown file, and the filename changes with the destination, such as 杭州_3日游_MCP.md. The repository also includes Chinese and English sample outputs under smart-travel-planner/examples/.

Files contain:

📅 Detailed daily itinerary arrangements
🏛️ Attraction introductions and recommendations
🍜 Food recommendations
💰 Budget allocation suggestions
🌤️ Weather information
🚇 Transportation suggestions

Project Structure

smart-travel-planner/
├── assets/
│   ├── picture7-1.png
│   └── picture7-2.png
├── docs/
│   ├── amd395-helloagents-smart-travel-planner-en.md
│   ├── amd395-helloagents-smart-travel-planner-zh.md
│   └── amd395-helloagents-smart-travel-planner-zh.pdf
├── examples/
│   ├── hangzhou-3-day-mcp.md
│   └── 杭州_3日游_MCP.md
└── travel_planner_mcp.py

Frequently Asked Questions

Q1: Model runs too slowly, what should I do?

Make sure to use AMD ROCm as the inference engine
Choose quantized versions of models (e.g., Q4_K_M)
Reduce model parameters (e.g., use 7B or 14B models)

Q2: Amap API call fails?

Check if the API Key is correctly configured
Confirm that the API Key's service type is "Web Service"
Check if the network connection is normal

Q3: How to switch to other map services?

Can be replaced with Baidu Maps, Tencent Maps, etc.
Adjust the service configuration inside MCPTool(...) in travel_planner_mcp.py
Replace the map provider API key and query parameters accordingly

Advanced Extensions

🌐 Multi-language Support: Add English, Japanese, and other language travel planning
🎨 UI Interface: Build web interface using Streamlit or Gradio
📊 Data Visualization: Generate itinerary maps and budget charts
🤝 Multi-person Collaboration: Support multiple people planning trips together
🔄 Real-time Updates: Dynamically adjust itinerary based on weather changes

Reference Resources

Build Your Own AI Travel Assistant with AMD GPU! 🗺️

Made with ❤️ by the hello-rocm community

chapter5

chapter6

smart-travel-planner

✈️ Intelligent Travel Planning Assistant - HelloAgents AMD ROCm Edition

Project Highlights

Technology Stack

Step 1: Environment Preparation

1.1 Install Python Dependencies

1.2 Obtain Amap API Key

Step 2: Local Model Deployment

Plan Selection

2.1 Using Ollama (Recommended)

2.2 Using LM Studio

2.3 Using Linglong "Smart Linglong Classmate"

Step 3: Configure the AMap API Key

3.1 Install uv

3.2 Configure the API Key

Step 4: Run the Intelligent Travel Planning Assistant

4.1 Download Project Code

4.2 Configure Model Endpoint

4.3 Run the Assistant

4.4 Usage Example

Step 5: View Generated Travel Plans

Project Structure

Frequently Asked Questions

Q1: Model runs too slowly, what should I do?

Q2: Amap API call fails?

Q3: How to switch to other map services?

Advanced Extensions

Reference Resources

✈️ Intelligent Travel Planning Assistant - HelloAgents AMD ROCm Edition ​

Project Highlights ​

Technology Stack ​

Step 1: Environment Preparation ​

1.1 Install Python Dependencies ​

1.2 Obtain Amap API Key ​

Step 2: Local Model Deployment ​

Plan Selection ​

2.1 Using Ollama (Recommended) ​

2.2 Using LM Studio ​

2.3 Using Linglong "Smart Linglong Classmate" ​

Step 3: Configure the AMap API Key ​

3.1 Install uv ​

3.2 Configure the API Key ​

Step 4: Run the Intelligent Travel Planning Assistant ​

4.1 Download Project Code ​

4.2 Configure Model Endpoint ​

4.3 Run the Assistant ​

4.4 Usage Example ​

Step 5: View Generated Travel Plans ​

Project Structure ​

Frequently Asked Questions ​

Q1: Model runs too slowly, what should I do? ​

Q2: Amap API call fails? ​

Q3: How to switch to other map services? ​

Advanced Extensions ​

Reference Resources ​

✈️ Intelligent Travel Planning Assistant - HelloAgents AMD ROCm Edition

Project Highlights

Technology Stack

Step 1: Environment Preparation

1.1 Install Python Dependencies

1.2 Obtain Amap API Key

Step 2: Local Model Deployment

Plan Selection

2.1 Using Ollama (Recommended)

2.2 Using LM Studio

2.3 Using Linglong "Smart Linglong Classmate"

Step 3: Configure the AMap API Key

3.1 Install uv

3.2 Configure the API Key

Step 4: Run the Intelligent Travel Planning Assistant

4.1 Download Project Code

4.2 Configure Model Endpoint

4.3 Run the Assistant

4.4 Usage Example

Step 5: View Generated Travel Plans

Project Structure

Frequently Asked Questions

Q1: Model runs too slowly, what should I do?

Q2: Amap API call fails?

Q3: How to switch to other map services?

Advanced Extensions

Reference Resources