Create an assistant - Pinecone Docs

An assistant answers questions grounded in the files you give it. Create one with a name and, optionally, instructions, metadata, and a region, then upload files and chat with it.

Create an assistant with the API or an SDK

To create an assistant, call create with a name and, optionally, instructions, metadata, and a region:

name: A name for the assistant, unique within the project.
instructions: A directive the assistant applies to every response it gives.
metadata: Optional key-value pairs, up to 16 KB, that help you organize assistants.
region: Where the assistant is deployed, us (default) or eu. You can’t change it later.

from pinecone import Pinecone

pc = Pinecone(api_key="YOUR_API_KEY")

assistant = pc.assistants.create(
    name="example-assistant",
    instructions="Use American English for spelling and grammar.",
    metadata={"team": "customer-support", "version": "1.0"},
    region="us",
)

import { Pinecone } from '@pinecone-database/pinecone';

const pc = new Pinecone({ apiKey: 'YOUR_API_KEY' });

const assistant = await pc.assistants.create({
  name: 'example-assistant',
  instructions: 'Use American English for spelling and grammar.',
  metadata: { team: 'customer-support', version: '1.0' },
  region: 'us',
});

PINECONE_API_KEY="YOUR_API_KEY"

curl "https://api.pinecone.io/assistant/assistants" \
  -H "Api-Key: $PINECONE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "name": "example-assistant",
  "instructions": "Use American English for spelling and grammar.",
  "metadata": {"team": "customer-support", "version": "1.0"},
  "region":"us"
}'

Instructions (maximum size 16 KB) are included in every chat API call. Longer instructions increase input token costs for each request and consume more of the LLM’s context window, reducing available space for retrieved context and conversation history.

You can also create an assistant in the Pinecone console.

Check when the assistant is ready

The Python SDK waits until the assistant is ready before it returns. The Node.js SDK and the API return right away, while the assistant’s status is still Initializing. To find out when it’s ready, get the status of the assistant.