Set up AI phone agents for customer calls
Implement Bland AI to automate calls and improve prospect conversion rates.
2024-12-20
Traditional call and support systems are often costly to maintain and limited in their capabilities, leaving businesses struggling to engage effectively with customers at scale. Automating customer calls with AI can significantly boost your sales team's efficiency and conversion rates.
Bland AI offers a solution to these challenges, providing a cost-effective and flexible platform for creating intelligent, AI-powered phone interactions.
This tutorial guides you through the process of leveraging Bland AI to transform your phone-based outreach. You'll learn how to craft conversations that adapt in real-time, understand context, and respond naturally to various scenarios - all without the need for extensive infrastructure or ongoing maintenance costs.
You'll need:
• Bland AI account
• CSV file with customer data
• ChatGPT (for prompt creation)
Let’s get to it.
Step 1: Create a Bland AI account and prepare data
Head to bland.ai and click “Get started for free" to sign up. After signing up, you'll be redirected to the dashboard.
.webp)
Prepare a CSV file containing customer information. The first row should be the headers, and column names should only use hyphens (-) or underscores (_). The phone number column must be labelled "phone_number".
To ensure proper formatting, download a sample CSV template from Bland AI. Two options are available:
- Starter CSV: A simplified template for basic data entry
- Complex CSV: An expanded template with additional fields for more detailed information
After signing up, you can access and download both CSV files by visiting here.
.webp)
Step 2: Create an AI agent prompt
Use ChatGPT to generate a customized prompt for your Bland AI agent. This prompt should cover comprehensive knowledge of your business to prevent AI hallucinations. Use the following prompt with ChatGPT:
Create a prompt for a Bland AI phone call agent using the following structure: Call Agent Name: (your desired agent name here)
1. Clear Goal and Routine:
Primary objective: [insert_main_call_objective]
Call flow:
• Step 1: [describe_step]
• Step 2: [describe_step]
• Step 3: [describe_step]
How to handle: [scenario_1], [scenario_2]
2. Background Information:
AI's role: [insert_ai_role]
Company/product: [briefly_describe_company_and_offering]
Typical customer: [describe_target_audience]
3. Example Dialogue:
Write a brief conversation demonstrating:
• Call introduction
• Handling objection: [insert_common_objection]
• Achieving call objective
4. Additional Instructions:
Tone: [describe_desired_tone]
Key phrases: [phrase_1], [phrase_2], [phrase_3]
Personalization: Use [specific_customer_data] to tailor the conversation
Ensure the final Bland AI prompt is strictly under 250 words, creating a natural, goal-oriented conversation.
.webp)
.webp)
Step 3: Set up bulk calls
On the Bland AI dashboard, click "Send Bulk Calls" to access the batch calling page.
.webp)
Configure your batch call settings:
- Enter your batch label
- Upload your CSV data file
.webp)
- Paste the generated prompt from ChatGPT
- Leave the Pathway ID as is (we're not covering conversation pathways in this tutorial)
.webp)
- Choose an AI model:
- Base: Best for following scripts and procedures
- Enhanced: Faster and more conversational, ideal for objective-based prompts
- Turbo: Fastest latency, but with limited capabilities
.webp)
- Set max call duration
- Select language and voice
- Leave temperature and interruption threshold as default
Step 4: Launch your batch calls
Click "Send Batch" in the top right corner to initiate your automated calls. You can toggle the "Test Call" option to make a single test call before starting the batch.
.webp)
This tutorial was created by Tanmay.