Introduction

We’re excited to announce that we just launched a Bland AI Zapier integration! Going forward, Bland customers can access all the functionality of our API through Zapier’s no-code platform.

‍

In this guide, we’ll show you how to access the Bland AI Zapier integration, create a custom workflow, and build your own AI appointment scheduler. During the process, you’ll learn how to configure your inbound phone call prompt, parse key information from a phone call transcript, and create and invite customers to a calendar event.

‍

Let’s get started!

Configuring your AI phone number

To start, you’ll need a Bland developer account. Visit our application portal and sign up. Next, visit the “Inbound Phone Numbers” page.

‍

Note: if you don’t already own an inbound phone number, you’ll have to purchase one.

Crafting your phone number’s prompt

For this example, we’ll build an appointment scheduler for “Dr. Linda Marvin’s Private Practice”. Copy/paste this prompt into the Custom prompt field:

‍

“You are Jared, a scheduling assistant at Dr. Linda Marvin's private practice. When someone calls to schedule an appointment, get their name and email. Repeat the email back to the person. Also figure out the specific date and time when the caller would like to schedule. The day of the week doesn't work - you need the exact date. Once you have all the required information end the phone call.”

Now you’re ready to set up Zapier.

Creating a Zap with Bland AI

Sign up for Zapier and then copy our pre-built appointment scheduler template here.

Initiating the phone call trigger

Click the webhook trigger block.

Press continue until you see your “Webhook URL”.

Copy the URL, return to the “Inbound Phone Numbers” page, and paste it into the Webhook field.

Note: Make sure you REMOVE the https:// from the start of your webhook because Bland pre-pends it by default.

‍

Testing the phone call trigger

Call your inbound phone number.

Revisit your Zapier editor and select “Test trigger”.

A phone call record should appear (if not, retest).

Select “Continue with Selected Record”.

‍

Parsing appointment details from the phone call

After receiving our phone call transcript from Bland AI, we must parse the following information from the phone call to schedule an appointment:

‍

1. The appointment start time
   
2. The appointment end time
   
3. The customer’s email address (to send them a calendar invitation)

‍

To parse the relevant details, we’ll start by cleaning the transcript. In this first step, we’ll ask GPT to collect all the relevant information from the transcript. In the subsequent steps, we’ll ask GPT to parse only those specific facts.

Cleaning the transcript

Click the first OpenAI action.

‍

Connect your OpenAI account (sign up here if you don’t already have one).

‍

Enter the following prompt: “Summarize all the appointment information. Parse the specific start time, end time, and the email address of the participant from this transcript:”

‍

From the data dropdown choose the Catch webhook, and then the “Transcript Text” attribute and append it to the prompt.

‍

Select continue, then test.

‍

Parsing the start time

Select the second OpenAI block

‍

Enter the following prompt: “Return the start time of the appointment. Only return the exact start time and date. No other text. ONLY THE START TIME AND DATE:”

‍

Append the OpenAI “response” attribute from the second step to the prompt.

‍

Select continue, then test.

‍

Getting the end time

Select the third OpenAI action.

‍

Use the following prompt: “Return the end time of the appointment. Only return the exact end time and date. No other text. ONLY THE END TIME AND DATE: [insert OpenAI “Response” attribute from 2nd step]. Note: if no end time exists, return ONLY THE START TIME AND DATE plus one hour.”

‍

Pulling the email‍

Select the fourth OpenAI action.

‍

Use the following prompt: “Parse any emails found in this transcript in comma separated format. Make sure to format the email correctly. E.g. replace "at" with "@". Here is the transcript: [insert OpenAI “Response” attribute from 2nd step]. ONLY RETURN EMAILS. NO OTHER TEXT.”

‍

Creating the calendar event

Select the Google Calendar block and connect your account.

‍

For the summary, provide the event title “Appointment with Dr. Linda Marvin”.

‍

For the location, enter whatever you prefer - e.g. “271 Battery St #650, San Francisco, CA 94111”.

‍

For the start date and time enter the response from the 2nd OpenAI block.

‍

For the end date and time enter the response from the 3rd OpenAI block.

‍

For the attendees enter the response from the 4th OpenAI block.

‍

Testing your Bland AI Zapier Integration

Before testing, publish your zap. Then call your inbound phone number and schedule a fake appointment.

‍

Tell the AI what your name and email are and when your preferred appointment time is. Then hang up the phone and check your calendar to see if the event is updated.

‍

Note: sometimes it takes Zapier a moment to complete the workflow. You can see the Zap’s status by visiting the history tab.

‍

Conclusion

We built the Bland AI Zapier integration to provide users a no-code option for sending AI phone calls.

‍

We'll continue creating additional templates and guides to enable more no-code workflows.

‍

Thanks for building on Bland!