Skip to main content

Build a Generative AI Travel Planner with Flutter, Firebase, and Google AI

Google for DevelopersApril 16, 20255 min17,434 views
17 connections·22 entities in this video→

Agentic Travel Planning App

  • πŸ’‘ An AI-powered travel planning app is demonstrated, built using Flutter, Vertex AI, Gemini, and the Places API.
  • 🎯 The app allows users to describe their desired vacation via text, image, or voice, with AI generating travel suggestions.
  • πŸ—ΊοΈ Suggestions are sourced from a database pre-generated using Gemini and Imagen.

Prototyping with Firebase Studio

  • πŸ› οΈ Firebase Studio is used for easy coding and integration of Google products.
  • 🎬 Two video walkthroughs showcase the app: one with a simple search interface and pre-generated results, and another with an AI-assisted multimodal search combining text and images.

Technology Stack and Integrations

  • πŸ”— Genkit framework provides AI capabilities, enabling agents to call tools for interacting with the environment, such as searching for nearby restaurants.
  • ☁️ A fully managed PostgreSQL database powered by Cloud SQL is integrated for scalable and secure data storage.
  • πŸ“Š Firebase Data Connect offers schema and query management using GraphQL, integrating with Firebase Authentication and enabling vector similarity search with queries like get nearest places.
  • πŸ“ The Places API is integrated via Genkit's tool calling to gather detailed location information and enhance recommendations.

Personalized Trip Generation with Gemini and RAG

  • ✈️ Users can import dream trips using text, speech-to-text, or images.
  • 🧠 Gemini API with Retrieval Augmented Generation (RAG) is leveraged to provide tailored trip overviews.
  • πŸ—„οΈ Firebase Data Connect and embeddings are used for nearest neighbor search, and Gemini and Imagen are used for pre-generating trip suggestions and images.

Flutter App Communication and Deployment

  • πŸ“± The Flutter application communicates with Firebase Genkit, defining schemas for validating LLM outputs and ensuring reliable data consumption.
  • πŸ“ Zod is used for defining input and output schemas for validation, and dotPrompt is a Genkit plug-in for crafting effective prompts.
  • πŸš€ The Flutter application is deployed to Firebase Hosting, which integrates seamlessly with modern web frameworks for secure and managed server environments.
Knowledge graph22 entities Β· 17 connections

How they connect

An interactive map of every person, idea, and reference from this conversation. Hover to trace connections, click to explore.

Hover Β· drag to explore
22 entities
Chapters3 moments

Key Moments

Transcript18 segments

Full Transcript

Topics15 themes

What’s Discussed

Generative AIAgentic ApplicationsFlutterFirebaseVertex AIGeminiImagenPlaces APIGenkitFirebase StudioCloud SQLFirebase Data ConnectGraphQLRetrieval Augmented Generation (RAG)Firebase Hosting
Smart Objects22 Β· 17 links
ProductsΒ· 10
ConceptsΒ· 10
MediasΒ· 2