How to Use OpenAI Assistants: Simple Guide with Examples

To use OpenAI assistants, you send a prompt or message to the assistant API and receive a response generated by the AI model. You interact by calling the assistant with your input text and processing its output to build chatbots or automation.

Syntax

The basic syntax to use an OpenAI assistant involves calling the assistant's chat.completions.create() method with a model name and a list of messages. Each message has a role (like user or assistant) and content (the text).

• model: The AI model to use (e.g., gpt-4o-mini).
• messages: A list of conversation turns with roles and texts.
• response: The assistant's reply text.

javascript

const openai = new OpenAI();

const response = await openai.chat.completions.create({
  model: "gpt-4o-mini",
  messages: [
    { role: "user", content: "Hello, who are you?" }
  ]
});

console.log(response.choices[0].message.content);

Output

Hello! I am an AI assistant here to help you with your questions.

Example

This example shows how to send a message to the OpenAI assistant and print its reply. It demonstrates a simple chat interaction where the user asks a question and the assistant responds.

javascript

import OpenAI from "openai";

async function runAssistant() {
  const openai = new OpenAI();

  const completion = await openai.chat.completions.create({
    model: "gpt-4o-mini",
    messages: [
      { role: "user", content: "What is the weather like today?" }
    ]
  });

  console.log("Assistant reply:", completion.choices[0].message.content);
}

runAssistant();

Output

Assistant reply: I'm sorry, I can't check real-time weather, but you can look it up online or use a weather app.

Common Pitfalls

Common mistakes when using OpenAI assistants include:

• Not specifying the model correctly, causing errors or unexpected results.
• Sending messages without the proper role field, which confuses the assistant.
• Ignoring the asynchronous nature of the API calls, leading to unhandled promises.
• Not handling API errors or rate limits gracefully.

Always check the API response structure and handle errors.

javascript

/* Wrong: Missing role in message */
const responseWrong = await openai.chat.completions.create({
  model: "gpt-4o-mini",
  messages: [
    { content: "Hello" } // Missing role
  ]
});

/* Right: Include role */
const responseRight = await openai.chat.completions.create({
  model: "gpt-4o-mini",
  messages: [
    { role: "user", content: "Hello" }
  ]
});

Quick Reference

• model: Choose the assistant model like gpt-4o-mini.
• messages: List of messages with roles user, assistant, or system.
• response: Access reply via response.choices[0].message.content.
• async: Use await to handle API calls.
• error handling: Always try/catch API calls.

Key Takeaways

Always specify the model and message roles correctly when calling the assistant.

Use async/await to handle the assistant API calls smoothly.

Check the assistant's reply in response.choices[0].message.content.

Handle errors and API limits to avoid crashes.

Start with simple prompts and build conversation context with message history.