Gmail

In this section, we present how to connect Gmail accounts to MindsDB.

Gmail is a widely used and popular email service developed by Google.

By connecting your Gmail account to MindsDB, you can utilize various AI models available within MindsDB to summarize emails, detect spam, or even automate email replies.

​

Prerequisites

Before proceeding, ensure the following prerequisites are met:

1. Install MindsDB locally via Docker or use MindsDB Cloud.
2. To connect Gmail to MindsDB, install the required dependencies following this instruction.
3. Install or ensure access to Gmail.

​

Connection

The required arguments to establish a connection are as follows:

• credentials_file local path to the credentials.json or credentials_url in case your file is uploaded to s3. Follow the instructions below to generate the credentials file.
• scopes define the level of access granted. It is optional and by default it uses ’https://…/gmail.compose’ and ’https://…/gmail.readonly’ scopes.

In order to make use of this handler and connect the Google Calendar app to MindsDB, the following syntax can be used:

CREATE DATABASE mindsdb_gmail
WITH ENGINE = 'gmail',
parameters = {
    "credentials_file": "mindsdb/integrations/handlers/gmail_handler/credentials.json",
    "scopes": ['https://.../gmail.compose', 'https://.../gmail.readonly', ...]
};

Or, you can also connect by giving the credentials file from an s3 pre signed url. To do this you need to pass in the credentials_file parameter as a pre signed url. For example:

CREATE DATABASE mindsdb_gmail
WITH ENGINE = 'gmail',
parameters = {
    "credentials_url": "https://s3.amazonaws.com/your_bucket/credentials.json?response-content-disposition=inline&X-Amz-Security-Token=12312...",
    -- "scopes": ['SCOPE_1', 'SCOPE_2', ...] -- Optional scopes. By default 'https://.../gmail.compose' & 'https://.../gmail.readonly' scopes are used
};

​

Usage

This creates a database called mindsdb_gmail. This database ships with a table called emails that we can use to search for emails as well as to write emails.

Now you can use your Gmail data, like this:

• searching for email:

  SELECT *
  FROM mindsdb_gmail.emails
  WHERE query = 'alert from:*@google.com'
  AND label_ids = "INBOX,UNREAD"
  LIMIT 20;
  

• writing emails:

  INSERT INTO mindsdb_gmail.emails (thread_id, message_id, to_email, subject, body)
  VALUES ('187cbdd861350934d', '8e54ccfd-abd0-756b-a12e-f7bc95ebc75b@Spark', 'test@example2.com', 'Trying out MindsDB',
          'This seems awesome. You must try it out whenever you can.');
  

​

Example 1: Automating Email Replies

Now that we know how to pull emails into our database and write emails, we can make use of OpenAPI engine to write email replies.

First, create an OpenAI engine, passing your OpenAI API key:

CREATE ML_ENGINE openai_engine
FROM openai
USING
    openai_api_key = 'your-openai-api-key';

Then, create a model using this engine:

CREATE MODEL mindsdb.gpt_model
PREDICT response
USING
    engine = 'openai_engine',
    max_tokens = 500,
    api_key = 'your_api_key', 
    model_name = 'gpt-3.5-turbo',
    prompt_template = 'From input message: {{body}}\
                        by from_user: {{sender}}\
                        In less than 500 characters, write an email response to {{sender}} in the following format:\
                        Start with proper salutation and respond with a short message in a casual tone, and sign the email with my name mindsdb';

​

Example 2: Detecting Spam Emails

You can check if an email is spam by using one of the Hugging Face pre-trained models.

CREATE MODEL mindsdb.spam_classifier                           
PREDICT PRED                           
USING
  engine = 'huggingface',              
  task = 'text-classification',        
  model_name = 'mrm8488/bert-tiny-finetuned-sms-spam-detection', 
  input_column = 'text_spammy',        
  labels = ['ham', 'spam'];

Then, create a view that contains the snippet or the body of the email.

CREATE VIEW mindsdb.emails_text AS(
    SELECT snippet AS text_spammy
    FROM mindsdb_gmail.emails
);

Finally, you can use the model to classify emails into spam or ham:

SELECT h.PRED, h.PRED_explain, t.text_spammy AS input_text
FROM mindsdb.emails_text AS t
JOIN mindsdb.spam_classifier AS h;