Agentic AI Demo with n8n, Ollama, and SearXNG

This repository provides a multi-container Docker setup for an Agentic AI demonstration. It integrates the n8n workflow automation tool, a self-hosted Ollama large language model, and the SearXNG private metasearch engine.

Download the demo repository

git clone https://github.com/AmpereComputingAI/ampere-ai-agents.git
cd ampere-ai-agents
git checkout <latest-release>

Prerequisites

Install Docker and Compose

Ensure that Docker Engine and Docker Compose are installed on your system. Follow the official instructions from the links below:
Docker Engine: https://docs.docker.com/engine/install/ubuntu/
Docker Compose: https://docs.docker.com/compose/install/linux/

Open the Required Ports

The application requires several ports to be accessible. Use the following commands to open them on a firewall like firewalld:

sudo firewall-cmd --zone public --permanent --add-port 5678/tcp
sudo firewall-cmd --zone public --permanent --add-port 8081/tcp
sudo firewall-cmd --zone public --permanent --add-port 11434/tcp
sudo firewall-cmd --reload

Start the Demo

Run the following command from the root of the project directory to start all the services in the background.

./start-app.sh

You can verify that all containers are running correctly:

$ docker ps
CONTAINER ID   IMAGE                                                 COMMAND                  CREATED          STATUS                    PORTS                                       NAMES
1a2b3c4d5e6f   ghcr.io/amperecomputingai/ampere-ai-agents:0.1.1      "docker-entrypoint.s…"   15 seconds ago   Up 14 seconds             0.0.0.0:5678->5678/tcp                      agentic_ai_demo_service
2b3c4d5e6f7g   docker.io/searxng/searxng:latest                      "/sbin/entrypoint.sh"    15 seconds ago   Up 14 seconds             0.0.0.0:8081->8080/tcp                      searxng
3c4d5e6f7g8h   ghcr.io/amperecomputingai/ollama-ampere:1.0.0-ol9     "bash -c 'ollama ser…"   15 seconds ago   Up 14 seconds (healthy)   0.0.0.0:11434->11434/tcp                    ollama_for_agent_service

Accessing the Service

Once the containers are running, you can access the Agentic AI from your web browser.

http://<server-IP>:5678

This is the main UI for building and managing your AI-powered workflows.

How to Use the Application

Automated Model Download

Upon starting, the Ollama service will automatically download the llama3.2:1b model. This process may take a few minutes depending on your internet connection. Once downloaded, the model is immediately available to be used within your n8n workflows.

Building Workflows

The core of this demo is the agentic_ai_demo_service, which runs n8n. You can build complex, agent-like workflows that can:

• Receive input via webhooks.
• Process logic and data.
• Query the Ollama language model for intelligent responses or analysis.
• Use the SearXNG service to perform web searches for real-time information.

For detailed instructions on how to create workflows, please refer to the official n8n documentation:

• https://docs.n8n.io/

Stop the Demo

To stop and remove all the containers and associated networks, run the following script:

./stop-app.sh

Services Overview

agentic_ai_demo_service: The core workflow automation platform powered by n8n.
ollama_for_agent_service: Provides local LLM inference via the Ollama API.
searxng: A private, hackable metasearch engine to provide workflows with web search capabilities.

License

This project is licensed under the MIT License. See the LICENSE file for details.

References

• https://n8n.io/
• https://github.com/ollama/ollama
• https://github.com/searxng/searxng