Upstash AI Search

Note

This project is in GA Stage.

The Upstash Professional Support fully covers this project. It receives regular updates, and bug fixes. The Upstash team is committed to maintaining and improving its functionality.

It is a connectionless (HTTP based) AI Search client and designed for:

• Serverless functions (AWS Lambda ...)
• Cloudflare Workers
• Next.js, Jamstack ...
• Client side web/mobile applications
• WebAssembly
• and other environments where HTTP is preferred over TCP.

Warning

This page is about Upstash Search, a standalone, fully managed AI search product with built-in semantic search, full-text search, and reranking.

It is not the same as Upstash Redis Search — a search extension built on Tantivy, available only on Upstash Redis and separate from the Redis Search API. For that, see the Upstash Redis Search docs.

Quick Start

Install

Node.js

npm install @upstash/search

Create Database

Create a new database on Upstash

Basic Usage:

import { Search } from "@upstash/search";

type Content = {
  title: string;
  genre: "sci-fi" | "fantasy" | "horror" | "action";
  category: "classic" | "modern";
};

type Metadata = {
  director: string;
};

// Initialize Search client
const client = new Search({
  url: "<UPSTASH_SEARCH_REST_URL>",
  token: "<UPSTASH_SEARCH_REST_TOKEN>",
});

// Create or access a index
const index = client.index<Content, Metadata>("movies");

// Upsert data into the index
await index.upsert([
  {
    id: "star-wars",
    content: { title: "Star Wars", genre: "sci-fi", category: "classic" },
    metadata: { director: "George Lucas" },
  },
  {
    id: "inception",
    content: { title: "Inception", genre: "action", category: "modern" },
    metadata: { director: "Christopher Nolan" },
  },
]);

// Fetch documents by IDs
const documents = await index.fetch({
  ids: ["star-wars", "inception"],
});
console.log(documents);

// AI search with reranking:
const searchResults = await index.search({
  query: "space opera",
  limit: 2,
  reranking: true,
});
console.log(searchResults);

// AI search without reranking:
const searchResults = await index.search({
  query: "space opera",
  limit: 2,
});
console.log(searchResults);

// AI search with only semantic search
const searchResults = await index.search({
  query: "space opera",
  limit: 2,
  semanticWeight: 1,
});

// AI search with only full-text search
const searchResults = await index.search({
  query: "space opera",
  limit: 2,
  semanticWeight: 0,
});

// AI search with full-text search and sematic search
// combined with equal weights
const searchResults = await index.search({
  query: "space opera",
  limit: 2,
  semanticWeight: 0.5,
});

// AI search without input enrichment
const searchResults = await index.search({
  query: "space opera",
  limit: 2,
  inputEnrichment: false,
});

// AI search without reranking:
const searchResults = await index.search({
  query: "space opera",
  limit: 2,
});
console.log(searchResults);

// AI search with filter:
const searchResults = await index.search({
  query: "space",
  limit: 2,
  filter: "category = 'classic'",
});

// Delete a document by ID
await index.delete({
  ids: ["star-wars"],
});

// Search within a document range
const { nextCursor, documents: rangeDocuments } = await index.range({
  cursor: 0,
  limit: 1,
  prefix: "in",
});
console.log(rangeDocuments);

// Reset the index (delete all documents)
await index.reset();

// Get index and namespace info
const info = await search.info();
console.log(info);