This page helps you get started with implementing generative AI features in yourapp. It describes the features and integrations for Firestore that involvegenerative AI.

Quickstart for vector search withCloud Firestore

Creating innovative AI-powered solutions for use cases such as productrecommendations and chatbots often requires vector similarity search, or vectorsearch for short. You can perform vector search on Firestore data without thehassle of copying data to another vector search solution, maintainingoperational simplicity and efficiency.

The core workflow for vector search inCloud Firestore consists of 4 steps.

Understand vector search fully in our blog post

looks_oneGeneratevector embeddings

The first step in utilizing vector search is to generate vector embeddings.Embeddings are representations of different kinds of data like text, images,and video that capture semantic or syntactic similarities between the entitiesthey represent. Embeddings can be calculated using a service, such as theVertex AI text-embeddings API.

looks_twoStoreembeddings in Firestore

Once the embeddings are generated you can store them in Firestore using one ofthe supported SDKs. Here is what that operation looks like in the NodeJS SDK:

constdb=newFirestore();letcollectionRef=db.collection("beans");awaitcollectionRef.add({name:"Kahawa coffee beans",type:"arabica",description:"Information about the Kahawa coffee beans.",embedding_field:FieldValue.vector([0.1,0.3,...,0.2]),// a vector with 768 dimensions});

looks_3Createa vector index

The next step is to create a Firestore KNN vector index where the vectorembeddings are stored. During the preview release, you will need to create theindex using thegcloud command line tool.

looks_4Performthe vector search

Once you have added all the vector embeddings and created the vector index, youare ready to run the search. You will then utilize thefind_nearest call on acollection reference to pass the query vector embedding with which to comparethe stored embeddings and to specify the distance function you want to utilize.

Once again, explore the workflow and more use cases in ourblog post.

Solution: vector search

Summary: Store and query vector embeddings.

Use case: This feature is used by the other tools and features.

See the guide for vector search

Solution: extension for vector search with Firebase

Summary: Use the Firebase extension to automatically embed and query yourFirestore documents with the vector search feature.

Use case: Perform automatic vector search in your Firebase projects.

Review the extension description

Solution: LangChain integrations

Summary: Use Firestore as a vector store, document loader, or chat messagehistory source for LangChain.

Use case: Build generative AI applications or retrieval-augmented generation(RAG) workflows.

See the guide for LangChain

Solution: Genkit

Summary: Genkit is an open source framework that helps you build,deploy, and monitor production-ready AI-powered apps.

Use case: Use Genkit andCloud Firestore to create apps that generatecustom content, use semantic search, handle unstructured inputs, answerquestions with your business data, and much more!

See the Genkit documentation