OpenAI + Slack = The Perfect Pair for Smart Assistants!

Create your own AI Powered Slack Bot Application

Photo by Austin Distelon Unsplash

In today’s fast-changing digital world, communication and working together are essential and Slack has become a revolutionary platform for teams. But what if I told you that you can improve your Slack experience by connecting it with OpenAI APIs? Imagine having a Slack bot that not only understands your questions and replies to them but also has access to extensive knowledge and advanced language abilities. We can achieve this with the help of OpenAI’s powerful models.

This article will explore the exciting world of integrating OpenAI APIs in Slack bots, unlocking a world of possibilities for enhanced productivity, intelligent automation, and seamless workflows.

Building the Slack App using Bolt and NodeJS

We will be using NodeJS to create a Slack application that will turn into an AI-powered smart assistant that can be used as a Smart Virtual assistant, Code assistant, Language Translator, Content generator, and so on. We can create a Slack application using Java, Python, and NodeJS. You can find more about it here and here. I will be referring to this Slack document.

1. Create your Slack App.

Create a Slack App from scratch

2. Give it a name and select the workspace, under which you want to add the app. After creating the app, it will be added under the selected workspace.

Give a name and add the Slack app to a Workspace.

3. Next, go to Features > OAuth & Permissions screen. Scroll down to Bot Token Scopes to specify the OAuth scopes.

Add required scopes.

4. Now you will need to add an app-level token.
Go to Settings > Basic Information screen, scroll down to App-Level Tokens, and add an app-level token for your app.

Generate App level Token

After this point, you will be able to install the application to the Slack Workspace.

Add the Slack application to the Slack Workspace by going to Settings → Install Apps and then proceed with Install to Workspace.

Install the app to the Workspace.

Allow Slack app to access the workspace.

You can also add slash commands for specific defined operations. Let’s add a command to the Slack app. Slash Commands and proceed with Create New Command. You can find more about Slash Commands here.

Create a new Slash Command.

Creating the NodeJS App

Let’s create a NodeJS application that will be the backbone of the Slack app. After creating a node application, follow these steps.

1. Create a .env file at the project root directory to store all environment variables. We will be using the dotenv library for developer dependencies.

touch .env

2. We will need the following keys and tokens from OpenAI and Slack API. These will be used to configure the Slack application and OpenAI authentication.

OPENAI_API_KEY = OPEN_API_KEY : https://platform.openai.com/account/api-keys
TOKEN =  TOKEN : Find in the OAuth  & Permissions tab
SIGNING_SECRET = SIGNING_SECRET: Find in Basic Information Tab
APP_TOKEN = APP_LEVEL_TOKEN: Find in Basic Information Tab

3. Install @slack/bolt. It is a NodeJS library that helps to connect with desired Slack applications. Execute the following command in the project root directory.

$ npm install @slack/bolt

4. Create an app by calling the constructor, a top-level export.

const { App } = require('@slack/bolt');

const app = new App({
  signingSecret: process.env.SIGNING_SECRET,
  token: process.env.TOKEN,
  appToken: process.env.APP_TOKEN
});


(async () => {
  // Start your app
  await app.start();

  console.log('⚡️ Bolt app is running!');
})();

5. Now let’s add a message handler that will catch every text message the user sends to this Slack application.

app.message(async ({ message, say }) => {  
  try {        
    let cmd = message.text // The inputted parameters
    console.log(cmd);
  } catch (error) {
      console.log("err")
    console.error(error);
  }
});

At this point, this application should be able to receive the text message sent by the user. You can check if we are receiving the text messages by logging those using console.log().

Interesting Part — Setting up Open AI

1. Install openai. Execute the following command in the project root directory.

$ npm install openai

2. Let’s configure Open-AI.library with the account secret key.

const { Configuration, OpenAIApi } = require("openai");

const configuration = new Configuration({
  apiKey: process.env.OPENAI_API_KEY,
});
const openai = new OpenAIApi(configuration);

3. Using the createChatCompletion() API, we can create the chat completion.

let prompt = "Hi, How are you?"
const completion = await openai.createChatCompletion({
      model: "gpt-3.5-turbo",
      messages: [
        {"role": "system", 
        "content": prompt}
      ],    
 });

4. Let’s combine the Slack message and openAI chat completion.

app.message(async ({ message, say }) => {  
  try {        
    let cmd = message.text
    const completion = await openai.createChatCompletion({
      model: "gpt-3.5-turbo",
      messages: [
        {"role": "system", 
        "content": cmd}
      ],    
    });

    say(completion.data.choices[0].message.content);
  } catch (error) {
      console.log("err")
    console.error(error);
  }
});

And that’s it. You have created the AI-powered Slack bot. Let’s run the NodeJS application and test the Slack app we just developed.

$ node app.js

Output :

[INFO]  socket-mode:SocketModeClient:0 Going to establish a new connection to Slack ...
⚡️ Bolt app is running!
[INFO]  socket-mode:SocketModeClient:0 Now connected to Slack

Here are the exciting responses got while interacting with an AI-powered Slack Assistant that we have created.

Smart Slack Application

Summary

We can create a Slack application to perform predefined tasks in our workspace. To add more power to the application, we used NodeJS. We created the application with NodeJS using the @slack/bolt library.

You can check for Bolt for Java & Bolt for Python, if you are willing to go for Java or Python instead of NodeJS.

After creating the Slack Application, we discovered how to integrate the OpenAI in the Node Application. We integrated the power of OpenAI in our NodeJS, and it made our Slack Application into a Smart Virtual Assistant.

References

OpenAI API

Hello World, Bolt ⚡️ - Building your very first app with Bolt

Slack | Bolt for JavaScript