Skip to main content

Convex

LangChain.js supports Convex as a vector store, and supports the standard similarity search.

Setup​

Create project​

Get a working Convex project set up, for example by using:

npm create convex@latest

Add database accessors​

Add query and mutation helpers to convex/langchain/db.ts:

convex/langchain/db.ts
export * from "langchain/util/convex";

Configure your schema​

Set up your schema (for vector indexing):

convex/schema.ts
import { defineSchema, defineTable } from "convex/server";
import { v } from "convex/values";

export default defineSchema({
  documents: defineTable({
    embedding: v.array(v.number()),
    text: v.string(),
    metadata: v.any(),
  }).vectorIndex("byEmbedding", {
    vectorField: "embedding",
    dimensions: 1536,
  }),
});

Usage​

tip

See this section for general instructions on installing integration packages.

npm install @langchain/openai @langchain/community

Ingestion​

convex/myActions.ts
"use node";

import { ConvexVectorStore } from "@langchain/community/vectorstores/convex";
import { OpenAIEmbeddings } from "@langchain/openai";
import { action } from "./_generated/server.js";

export const ingest = action({
  args: {},
  handler: async (ctx) => {
    await ConvexVectorStore.fromTexts(
      ["Hello world", "Bye bye", "What's this?"],
      [{ prop: 2 }, { prop: 1 }, { prop: 3 }],
      new OpenAIEmbeddings(),
      { ctx }
    );
  },
});

API Reference:

convex/myActions.ts
"use node";

import { ConvexVectorStore } from "@langchain/community/vectorstores/convex";
import { OpenAIEmbeddings } from "@langchain/openai";
import { v } from "convex/values";
import { action } from "./_generated/server.js";

export const search = action({
  args: {
    query: v.string(),
  },
  handler: async (ctx, args) => {
    const vectorStore = new ConvexVectorStore(new OpenAIEmbeddings(), { ctx });

    const resultOne = await vectorStore.similaritySearch(args.query, 1);
    console.log(resultOne);
  },
});

API Reference:

Help us out by providing feedback on this documentation page: