ChromaDB Has a Free API: The AI-Native Vector Database for RAG Applications

#ai #database #python #machinelearning

Chroma is an open-source vector database designed for AI applications. It makes building RAG (Retrieval-Augmented Generation) pipelines as simple as adding documents and querying.

Why Chroma Matters

RAG needs a vector store. Most options are complex (Pinecone, Weaviate) or limited (FAISS). Chroma is batteries-included: embed, store, search in 4 lines of code.

What you get for free:

4-line setup: create collection, add, query
Built-in embedding (no separate embedding service needed)
Document and metadata storage alongside vectors
Python and JavaScript clients
Runs embedded or as a server
Where filters for metadata
Multi-modal support

Quick Start

pip install chromadb

import chromadb

client = chromadb.Client()

# Create collection (auto-embeds with sentence-transformers)
collection = client.create_collection("docs")

# Add documents — Chroma embeds them automatically
collection.add(
    documents=[
        "AI is transforming healthcare with diagnostic tools",
        "Machine learning models predict stock market trends",
        "Neural networks process images for self-driving cars",
        "NLP enables chatbots to understand human language",
    ],
    metadatas=[
        {"topic": "healthcare"},
        {"topic": "finance"},
        {"topic": "automotive"},
        {"topic": "nlp"},
    ],
    ids=["doc1", "doc2", "doc3", "doc4"],
)

# Query — returns most similar documents
results = collection.query(
    query_texts=["How is AI used in medicine?"],
    n_results=2,
)
print(results["documents"])  # Healthcare doc comes first

RAG with LangChain

from langchain_chroma import Chroma
from langchain_openai import OpenAIEmbeddings, ChatOpenAI
from langchain.chains import RetrievalQA

# Create vector store
vectorstore = Chroma.from_documents(
    documents=docs,
    embedding=OpenAIEmbeddings(),
    persist_directory="./chroma_db",
)

# Create RAG chain
qa = RetrievalQA.from_chain_type(
    llm=ChatOpenAI(model="gpt-4o"),
    retriever=vectorstore.as_retriever(search_kwargs={"k": 5}),
)

answer = qa.invoke("What are the latest AI trends?")
print(answer)

Metadata Filtering

results = collection.query(
    query_texts=["AI applications"],
    n_results=5,
    where={"topic": {"$in": ["healthcare", "finance"]}},
)

# Complex filters
results = collection.query(
    query_texts=["machine learning"],
    where={
        "$and": [
            {"topic": {"$eq": "tech"}},
            {"year": {"$gte": 2024}},
        ]
    },
)

Persistent Storage

# Persist to disk
client = chromadb.PersistentClient(path="./chroma_data")
collection = client.get_or_create_collection("docs")

# Or run as a server
# chroma run --path ./chroma_data

JavaScript Client

import { ChromaClient } from "chromadb";

const client = new ChromaClient();
const collection = await client.createCollection({ name: "docs" });

await collection.add({
  ids: ["doc1", "doc2"],
  documents: ["AI in healthcare", "ML in finance"],
  metadatas: [{topic: "health"}, {topic: "finance"}],
});

const results = await collection.query({
  queryTexts: ["medical AI"],
  nResults: 2,
});

Links

Building RAG applications? Check out my developer tools on Apify or email spinov001@gmail.com for custom solutions.

DEV Community