Extracting information from emails is a powerful productivity booster. In this article, we will explain several ways to get in the information out from an email.

NOTE: It is important to note that Bitskout is not the best platform to general mailbox filtering. It is highly recommended that you already established a filter, for instance, a separate mailbox for orders or some rules.

Also, we recommend having one model per topic/email type. It is not recommended to create broad models as they will not perform well.

Now, typically there are two types of email content:

  • Structured where content is structured in a certain way

  • Unstructured where it is basically written as a normal human text

Structured Emails

Typically, emails with structured content are generated by tools. Their format rarely changes and they serve as notifications.

As you can see from the example the email has a certain structure - it has fields like Reason, Scenario, and Organization that are clearly separated from their values.

Here is another example:

We can clearly see the structure - there are fields like First Name, Last Name, and Email.

Hence, to extract structured information we can use the Forms model in Bitskout.

  1. Go to the Models page and add a new A.I. model.

  2. Specify the Model name and choose Data Extraction

  3. Then scroll down to the list of options and choose Extract Forms from the drop-down list:

  4. Once you've selected the Forms options, an extra screen with a description will appear:

  5. Scroll down and load an example. In this case, you can take a screenshot of your email or save it as a PDF to be able to load it into Bitskout:

  6. Once the file is loaded, you will need to press Refresh the list of fields to see what Bitskout detects.

  7. Then you will be able to select the fields that you need in your task:

    NOTE: In case some field is not extracted, then you can create a model that works with unstructured data (see the instruction below).

  8. Scroll down and click the checkbox to create the workflow automatically and then press Apply to save your model and create a new workflow.

The next step would be to create an output to write extracted data to the task columns.

Unstructured Emails

Unstructured emails are all other messages that are written as it is without any defined structure. Here is the example:

In this example email, you can see the request for Batman services. In a work environment, this email contains crucial information - a location, a contact name, and a phone number.

Let's check another example:

Even though it is a notification, this email has an unstructured format. There are two items of interest from this email - a contact email and the URL of the spreadsheet.

Let's configure the data extraction from such emails:

  1. Go to the Models page and add a new A.I. model.

  2. Choose Text Creation from the list of model types.

  3. Once you select the type, the screens with options will appear. Choose the option Extract Data from Unstructured Text:

  4. Configure model parameters - you can use the value that you see on the screen below:

  5. And now the important part - click on the Setup Examples button.

  6. The screen with examples setup will appear.

  7. Click Add Example:

  8. Add text into the text box and then add one by one the fields you want to extract. Add more examples if required. We recommend 4-5 examples.

    As an example, here is how the examples look like for Google Sheets:

  9. Once you are done with examples, press Close. Select the checkbox to automatically create a workflow and press Apply.

    The next step would be to create an output to write extracted data to the task columns.

    A demo video for monday.com

Did this answer your question?