AI,M 9 3

Summary

The video demonstrates how to build a workflow that reads customer records from Airtable, selects useful fields, and routes records according to a country condition. Although the presenter describes this as training a customer data store, the workflow primarily retrieves and transforms existing customer data rather than training a machine-learning model.

Tutorial and key concepts

1. Get sample customer data

   • Download a customer CSV containing fields such as customer ID, first and last name, company, city, country, phone number, email, subscription date, and website.
   • The presenter uses 5,000 records for the example.

2. Import the data

   • Google Sheets is shown as one option, but the presenter recommends Airtable because its records are easier to organize and manage.
   • Create an Airtable base and import the CSV. Excel files, Google Sheets, or pasted data can also be imported.

3. Build the workflow

   • Start with a manual trigger.
   • Add a customer data store action to retrieve all persons.
   • Connect the data store to Airtable and use Search records to fetch existing customer records. An Airtable API access token may be required if credentials are not already connected.
   • MongoDB and Supabase are mentioned as possible direct database alternatives, but they are not demonstrated.

4. Transform and filter records

   • Use a data transformation step to keep selected fields: full name (combining first and last name), email, phone number, and country.
   • Add an If condition that checks whether the country equals “Bangladesh.”
   • Send matching records down the true branch and records from other countries down the false branch. The presenter reports 45 Bangladesh records and 4,955 records from other countries, out of 5,000 total.

5. Route results for later automation

   • Further transform each branch to retain names and phone numbers, then label the outputs for Bangladeshi and foreign users.
   • The presenter explains that this separation could support later automation—for example, directing different groups to agents who can contact them in Bengali or English, or sending calls or SMS. These outreach actions are discussed as future uses; they are not performed in the tutorial.

Main speaker and source

A single presenter from the Ai Automation Mastery channel; no other speakers are identified.