This repository contains the source code for Google AI Studio's starter apps — a collection of small apps that demonstrate how Gemini can be used to create interactive experiences. These apps are built to run inside AI Studio, but the versions included here can run standalone using the Gemini API.

Spatial Understanding (live demo) is a simple demonstration of Gemini's 2D and 3D spatial understanding and reasoning capabilities. It was built with React.

This example should give you an idea of how to get started with spatial analysis using Gemini. Check out Prompt.tsx to see how bounding box parsing is implemented. To dive deeper into Gemini's spatial reasoning capabilities, check out this Colab notebook.

To develop locally, insert your Gemini API key in the .env file.

Video Analyzer (live demo) is a simple app that allows you to explore events within videos using Gemini. It was built with React.

This example shows how to get started with video analysis using function calling. Check out functions.js to see the function definition for this applet!

To develop locally, insert your Gemini API key in the .env file.

Map Explorer (live demo) is a simple app that allows you to explore a map using Gemini and the Google Maps API. It was built using Lit, and the Maps Embed API.

This example will give you an idea of how to get started with function calling. See function-declarations.js to find out more about how function calling is used to call the Maps Embed API here!

To develop locally, insert your Gemini API key where you see your_key_here in the script.js file.

• @bencobley
• @dmotz
• @grantcuster
• @hapticdata