Build Your Own AI Chat App with Google Gemini and Cloud Run

Large Language Models (LLMs) like Google’s Gemini offer powerful capabilities for creating interactive applications. This guide focuses on building a simple yet functional web-based chat application powered by the Gemini API and deploying it seamlessly using Google Cloud Run. This tutorial is ideal for developers exploring LLM integration or looking for practical ways to deploy AI features in their web projects.

What This Guide Covers

Follow this tutorial to construct a web chat interface that interacts directly with the Gemini API to provide conversational responses. The application will be containerized using Docker and deployed onto Google Cloud Run via Cloud Build and Artifact Registry.

Upon completion, you will have learned how to:

• Set up a Python web application interacting with the Gemini API.
• Containerize the application using Docker.
• Deploy the container to Cloud Run using Google Cloud tools.
• Gain insights into integrating LLMs into web applications.

Think of it as building your own helpful chat assistant, ready to deploy to the cloud.

Prerequisites

Before starting, ensure the following prerequisites are met:

• A Google Cloud project.
• The gcloud command-line interface installed and authenticated.
• Docker installed on your machine.
• The Vertex AI API enabled within your Google Cloud project.
• (Optional: Google Cloud Shell provides a pre-configured environment with most tools ready.)

Step 1: Obtain the Starter Code

Begin by cloning the pre-built Python Flask application template from its repository. Open your terminal and run:

git clone https://github.com/ChloeCodesThings/chat-app-demo
cd chat-app-demo

Inside the cloned directory, you’ll find these key files:

• app.py: Contains the Flask application routes and the logic for interacting with the Gemini API.
• index.html: Provides a basic user interface for the chat application.
• Dockerfile: Includes instructions for building the application’s container image.

Step 2: Build and Store the Docker Image using Cloud Build

Containerizing the application makes deployment consistent and reliable. First, define necessary environment variables in your terminal. Replace us-central1 with your preferred Google Cloud region if needed.

export PROJECT_ID=$(gcloud config get-value project)
export REGION=us-central1
export AR_REPO=chat-app-repo
export SERVICE_NAME=chat-gemini-app

Next, create a repository in Google Artifact Registry to store your Docker image:

gcloud artifacts repositories create $AR_REPO \
--repository-format=docker \
--location=$REGION

Finally, use Cloud Build to build the Docker image based on your Dockerfile and push it to the Artifact Registry repository you just created:

gcloud builds submit --tag $REGION-docker.pkg.dev/$PROJECT_ID/$AR_REPO/$SERVICE_NAME

Cloud Build automates the build process using the instructions in your Dockerfile.

Step 3: Deploy the Application to Cloud Run

With the container image built and stored in Artifact Registry, you can deploy it to Cloud Run, Google’s serverless platform for containerized applications:

gcloud run deploy $SERVICE_NAME \
  --image=$REGION-docker.pkg.dev/$PROJECT_ID/$AR_REPO/$SERVICE_NAME \
  --platform=managed \
  --region=$REGION \
  --allow-unauthenticated \
  --set-env-vars=GCP_PROJECT=$PROJECT_ID,GCP_REGION=$REGION

This command deploys your container image as a new Cloud Run service. The --allow-unauthenticated flag makes the service publicly accessible (for production, you’d typically implement authentication). Cloud Run will provide a URL for your deployed service, similar to `https://chat-gemini-app-xxxxxx.run.app`.

Step 4: Understanding the Gemini API Interaction

The core AI logic resides in app.py. Here’s a snippet showcasing the interaction with the Vertex AI SDK for Python, which facilitates communication with the Gemini API:

from vertexai.preview.language_models import ChatModel

def create_session():
    chat_model = ChatModel.from_pretrained("gemini-pro") # Or other available models like gemini-1.0-pro
    chat = chat_model.start_chat()
    return chat

def response(chat, message):
    result = chat.send_message(message)
    return result.text

This code initializes a chat session using a specified Gemini model (like gemini-pro). When a user submits a message through the web interface, the response function sends this message to the Gemini model via the established chat session and returns the model’s text response back to the user interface.

Test Your AI Chat App

Open the deployment URL provided by Cloud Run in your browser. Enter a prompt or question, for example:

“Explain Google Cloud Run in simple terms.”

You should receive a relevant, conversational response generated by the Gemini model running via your deployed application.

Expanding Your Project

This basic application serves as a great starting point. Consider these potential enhancements for future development:

• Improving the chat interface for a better user experience.
• Implementing session persistence to maintain chat history across interactions.
• Using system prompts to guide Gemini’s personality and response style.
• Securing the application endpoint using Identity Platform, API Gateway, or other authentication methods.

Summary

Congratulations! You have successfully built and deployed a chat application powered by Google Gemini on Cloud Run. This involved setting up a Flask app, containerizing it with Docker, utilizing Cloud Build and Artifact Registry, and deploying to a scalable serverless platform. This project provides a solid foundation for integrating powerful AI capabilities into your web applications.

---

Leverage AI Power with Innovative Software Technology

Integrating advanced AI like Google Gemini into business applications unlocks significant value, enhancing customer engagement and automating workflows. At Innovative Software Technology, we specialize in developing and deploying cutting-edge AI solutions tailored to your needs. Our expert team excels at seamless Gemini API integration, building scalable web applications, and managing robust Google Cloud Run deployments. Partner with Innovative Software Technology for expert AI application development and transform your operations with intelligent, high-performing custom software solutions on the cloud. We turn complex AI possibilities into tangible business results.