Inspiration
The first inspiration was the challenge prize itself. It motivated me to explore Google’s Gemini AI and think about how an AI agent could interact with users in a simple and engaging way. I wanted to create a demo that shows the experience of using a multimodal AI assistant while keeping the interface accessible and interactive.
What it does
Gemini Spark is an interactive web-based simulation of a Gemini-powered AI assistant. Users can log in, ask questions through a chat interface, and view their previous conversations in an organized “Old Chats” panel. The assistant provides responses that demonstrate how an AI agent can interact with users, while also linking to Vertex AI Studio where the real Gemini model can be explored. The goal is to show how an AI assistant could guide users and manage conversations in a clean and intuitive interface
How we built it
Gemini Spark was built as a lightweight web application. Technologies used: HTML5 for the structure of the interface CSS3 for styling the login page and chat UI JavaScript (ES6) for chat interaction and logic LocalStorage to store previous conversations in the browser Vertex AI Studio as a reference to the real Gemini model The project includes: A login interface A chat interface with a send arrow and Enter key support A chat history panel delayed responses to simulate AI processing behavior
Challenges we ran into
One of the biggest challenges was working around API request limits and billing verification for the Gemini API. Because of this, I needed to design a way to demonstrate the AI interaction without relying on constant live API calls. Another challenge was creating a user interface that feels realistic and responsive, including login behavior, chat flow, and old chat history.
Accomplishments that we're proud of
Designing a clean and interactive AI chat interface Implementing login flow and conversation history Simulating realistic AI interaction with delayed responses Creating a demo that clearly shows how an AI assistant could work in a real application
What we learned
Through this project I learned how to: Design a complete AI-style chat interface Improve user experience in conversational apps Simulate AI behavior using frontend logic Structure a project that demonstrates AI concepts clearly
What's next for Spark
Future improvements could include: Connecting the interface to the live Gemini API Adding multimodal inputs like images or audio Expanding the chat history into full conversation sessions Deploying the project as a public web application
Log in or sign up for Devpost to join the conversation.