Build a Slack Chatbot using MindsDB and OpenAI GPT Model: A Step-by-Step Guide

Tarun Chawla — Tue, 18 Jul 2023 03:13:34 +0000

The purpose of this tutorial is to guide you in creating an AI-powered, personalized chatbot by leveraging MindsDB's Slack Integration and OpenAI's GPT-4 Model.

To demonstrate this, we will build a Slack bot, @Whiz_Fizz, which will intelligently respond to user inquiries while maintaining a unique persona as a quirky magician 🪄 and space science expert 🪐. Let's explore how it responds.

Before Jumping more into it. Let's first see how to create a bot and connect it to our Slack Workspace.

To get started
Generating API Tokens
Implementation
Crafting the GPT-4 Model
Feeding Personality into Our Model
Let's Connect our GPT Model to Slack
Posting Messages using SQL
Automation
Conclusion

To get started :

Create an account on MindsDB cloud (if you don’t have one yet).
Create a Slack Account (if you don't already know about Slack.)
Go to your MindsDB SQL Editor

MindsDB is a famous open-source application used for simplifying the use of Machine Learning in modern-day applications without the need for any external tool by making forecasts on the Data available in your database. It lets you build any Generative AI application with minimal steps required.

If you already have the API token of a Slack Bot, feel free to skip the following steps.

Generating API Tokens

If you have a Slack Account, you have to generate the API token, please follow below instructions:

Here are the steps to generate the Slack API token:

Follow this Link and sign in with your Slack Account
Create a new App or select an existing App.
In the app settings, go to the OAuth & Permissions section.
Under the Bot Token Scopes section, add the necessary scopes for your application. You can add more later as well.
Install the bot to your workspace.
In the OAuth Tokens & Redirect URLs Section, copy the the Bot User OAuth Access Token (This is the API token which we need).
Open your Slack, in order to use the bot which we created, we have to add the bot into the channel where we want to use this.
- Go to the channel where you want to use the bot.
- Right Click on the channel and select View Channel Details.
- Select Integrations.
- Click on Add an App.
- You can see the name of the bot under the In your workspace Section, Go ahead and add the app to the channel.

Now, we can use the token from step 6 to initialize the Slack Handler in MindsDB.

Now, let's establish a connection to our Slack using MindsDB SQL Editor:

CREATE DATABASE mindsdb_slack 
WITH 
    ENGINE = 'slack', 
    PARAMETERS = { 
        "token": "<slack-bot-token>" 
    };

Please change the slack-bot-token with the token mentioned in Bot User OAuth Access Token.

This query will create a database called mindsdb_slack and channels table automatically.

Implementation

Here is how to retrieve the 10 latest messages from the channel:

SELECT *
FROM mindsdb_slack.channels
WHERE channel="<channel-name>"
LIMIT 10;

Please change the channel-name in the WHERE clause to the channel where, you added the bot in your Slack Workspace.

However, you can also retrieve messages in alphabetical order by using:

SELECT *
FROM mindsdb_slack.channels
WHERE channel="<channel-name>"
ORDER BY messages ASC
LIMIT 5;

By default, it retrieves by the order the messages were sent, unless specified ascending/descending.

Here is how to post messages:

INSERT INTO mindsdb_slack.channels (channel, message)
VALUES
    ("<channel-name>", "Hey MindsDB, Thanks to you! Now I can respond to my Slack messages through SQL Queries. 🚀 "),
    ("<channel-name>", "It's never been that easy to build ML apps using MindsDB!");

Whoops! Sent it by mistake, No worries, use this to delete a specific message

DELETE FROM mindsdb_slack.channels
WHERE channel = "<channel-name>" AND ts = "1688863707.197229";

Now, let's roll up our sleeves and start building the GPT-4 Model together.

Here's how we can do it:

1. Crafting the GPT-4 Model :

Generating a Machine Learning model with MindsDB feels like taking a thrilling elevator ride in Burj Khalifa (You don't realize, that you made it)!

Here gpt_model represents our GPT-4 Model.

(OpenAI API is in HIGH demand and is rate limited, you might experience some delays in getting responses.)

CREATE MODEL mindsdb.gpt_model
PREDICT response
USING
engine = 'openai',
max_tokens = 300,
-- api_key = '<your-api-token>',
model_name = 'gpt-3.5-turbo',
prompt_template = 'From input message: {{messages}}\
write a short response to the user in the following format:\
Hi, I am an automated bot here to help you, Can you please elaborate the issue which you are facing! ✨🚀 ';

If you are using MindsDB locally or using Docker or if you want to use your own OpenAI API Key, simply remove -- from api_key argument.

The critical attribute here is prompt_template where we tell the GPT model, how to respond to the questions asked by the user, feel free to change any values passed in the columns, from our example, {{messages}}, will be replaced by the WHERE clause provided in the query. Let's see how it works:

SELECT
  messages, response
FROM mindsdb.gpt_model
WHERE messages = 'Hi, can you please explain me more about MindsDB?';

2. Feeding Personality into Our Model

Alright, so the old model's replies were good. But hey, we can use some prompt template tricks to make it respond the way we want. Let's do some Prompt Engineering.

Now, let's make a model called whizfizz_model with a prompt template that gives GPT a wild personality that eludes a playful and magical aura. Imagine scientific knowledge with whimsical storytelling to create a unique and enchanting experience. We'll call him WhizFizz:

CREATE MODEL mindsdb.whizfizz_model
PREDICT response
USING
engine = 'openai',
max_tokens = 300,
-- api_key = 'your openai key, -- in MindsDB cloud accounts we provide a default key
model_name = 'gpt-4', -- you can also use 'text-davinci-003' or ''
prompt_template = 'From input message: {{messages}}\
write a short response in less than 40 words to some user in the following format:\
Hi there, WhizFizz here! <respond with a mind blowing fact about Space and describe the response using cosmic and scientific analogies, where wonders persist. In between quote some hilarious appropriate raps statements based on the context of the question answer as if you are a Physics Space Mad Scientist who relates everythign to the Universe and its strange theories. So lets embark on a journey, where science and magic intertwine. Stay tuned for more enchantment! ✨🚀 -- mdb.ai/bot by @mindsdb';

Let's test this in action:

SELECT
  messages, response
FROM mindsdb.whizfizz_model
WHERE messages = 'Hi, can you please explain me more about MindsDB?';

You see the difference! Now, I'm getting excited, let's try again.

SELECT
  messages, response
FROM mindsdb.whizfizz_model
WHERE messages = 'if a time-traveling astronaut had a dance-off with a black hole, what mind-bending moves would they showcase, and how would gravity groove to the rhythm?!';

3. Let's Connect our GPT Model to Slack!

The database has channels table. This table can be used to search for channels, messages, and timestamps, as well as to post messages into slack channels. These functionalities can also be done by using Slack API or Webhooks.

Let's query the user's question and see how our GPT model responds to it, by joining the model with the channels table:

SELECT
    t.channel as channel,
    t.messages as input_text, 
    r.response as message
FROM mindsdb_slack.channels as t
JOIN mindsdb.whizfizz_model as r
WHERE t.channel = '<channel-name>'
LIMIT 3;

4. Posting Messages using SQL

We want to respond to the user's questions by posting the output of our newly created WhizFizz Model. Let's post the message by querying and joining the user's questions to our model:

INSERT INTO mindsdb_slack.channels(channel, message)
  SELECT
    t.channel as channel,
    t.messages as input_text, 
    r.response as message
  FROM mindsdb_slack.channels as t
  JOIN mindsdb.whizfizz_model as r
  WHERE t.channel = '<channel-name>'

5. Let's automate this

We will CREATE JOB that means the query which we will give, will execute after some interval. Let's schedule a job every hour. The job will do the following: 1. Check for new messages 2. Generate an appropriate response 3. Post the response into the channel

Let's do it in a single SQL statement:

CREATE JOB mindsdb.gpt4_slack_job AS (
   -- insert into channels the output of joining model and new responses
  INSERT INTO mindsdb_slack.channels(channel, message)
  SELECT
    t.channel as channel,
    t.messages as input_text, 
    r.response as message
  FROM mindsdb_slack.channels as t
  JOIN mindsdb.whizfizz_model as r
  WHERE t.channel = '<channel-name>'
  LIMIT 10;
)
EVERY hour;

To check the jobs and jobs_history, we can use the following:

SELECT * FROM jobs WHERE name="gpt4_slack_job";
SELECT * FROM jobs_history WHERE name="gpt4_slack_job";

To stop the sheduled job, we can use the following:

DROP JOB gpt4_slack_job;

6. Conclusion

The tutorial demonstrated how to create an personalized AI-powered Slack chatbot using MindsDB and OpenAI's GPT-4 model. By following the step-by-step guide, you can build a unique persona for your chatbot, connect it to Slack, and automate its responses to user queries. This powerful combination of technologies enables the creation of engaging and interactive chat experiences for users in any Slack workspace.

DEV Community: Tarun Chawla