Skip to main content

Overview

Creating an agent in Graine AI is a streamlined 3-step process that takes approximately 5-10 minutes. You’ll configure the voice, behavior, and capabilities of your AI agent.

Agent Creation Process

1

Step 1: Voice Selection

Choose the voice and language for your agent

Select Template (Optional)

Start with a pre-built template:
  • Customer Support - Handle customer inquiries and support tickets
  • Sales Assistant - Qualify leads and schedule appointments
  • Appointment Booking - Schedule and manage appointments
  • Survey & Feedback - Collect customer feedback and surveys

Choose Voice & Language

Agent Name: Give your agent a descriptive namePrimary Language: Select from 10+ languages
  • English, Hindi, Tamil, Malayalam, Telugu
  • Bengali, Gujarati, Punjabi, Marathi, Kannada, Spanish
Secondary Language: Optional fallback for multilingual supportVoice Selection: Choose from multiple voice profiles
  • Male voices: Mahesh, Rahul, Abhilash
  • Female voices: Manisha, Vidya, Susheela, Anika, Anushka
Preview each voice before selecting to find the perfect match
2

Step 2: Configuration

Configure your agent’s knowledge and behavior ### Product Information (Required) Company Name - Enter your company or organization name - Used in the agent’s introduction Call Purpose - Clearly state the purpose of calls - Examples: “Sales outreach”, “Customer support”, “Appointment booking” Additional Information - Provide context about your business, products, or services - Include key information the agent should know - Be specific and comprehensive ### Frequently Asked Questions (FAQs) Add common questions and answers: - Click “Add FAQ” to create question-answer pairs - Each FAQ should address a common customer inquiry - Provide clear, accurate, and concise answers Best Practices: - Cover your most common inquiries - Keep answers concise (2-3 sentences) - Use natural, conversational language - Update regularly based on customer interactions ### Introduction / Greeting Message Write the opening message your agent will use: plaintext Example: "Hello! I'm calling from Acme Corporation. My name is Sarah, and I'm an AI assistant here to help you with scheduling your appointment. Is this a good time to talk?"
Make your greeting warm, professional, and clear about the call’s purpose

Custom Variables Personalize conversations dynamically: **Default

Variables**: - callee_name - Customer’s name - mobile_number - Phone number Adding Custom Variables: - Click “Add Variable” - Enter variable name (e.g., appointment_date, order_number) - Provide default value (optional) - Use in prompts as {variable_name} Example usage: plaintext "Hello {callee_name}, I'm calling about your appointment on {appointment_date} " ### Knowledge Base Connect external knowledge sources: 1. Navigate to “Knowledge Base” in the sidebar 2. Select from your uploaded collections 3. Connect desired knowledge bases to your agent 4. Agent will automatically reference these documents during calls
3

Step 3: Deploy

Launch your agent
  1. Click “Create Agent”
  2. Wait for generation (typically 10-30 seconds)
  3. Your agent is now ready to use!
Your agent will appear in your dashboard with a unique ID

Templates

Customer Support Template

Pre-configured for handling customer inquiries:
  • Professional greeting
  • FAQ structure for common issues
  • Escalation paths
  • Ticket creation capabilities

Sales Assistant Template

Optimized for lead qualification:
  • Qualifying questions
  • Appointment scheduling
  • Product information delivery
  • Follow-up coordination

Appointment Booking Template

Streamlined scheduling process:
  • Availability checking
  • Calendar integration
  • Confirmation handling
  • Reminder setup

Survey & Feedback Template

Designed for data collection:
  • Structured question flow
  • Rating scales
  • Open-ended responses
  • Thank you messaging

Best Practices

Templates provide proven conversation flows. Customize them rather than starting from scratch.
  • Aim for 2-3 sentence answers - Use simple language - Cover 80% of common questions - Add more over time based on actual calls
Your greeting sets the tone. Test multiple versions to find what works best for your audience.
Instead of var1, use appointment_date or order_number for clarity.

After Creation

Once created, you can:

Test Your Agent

Use web calling to test immediately (Coming Soon)

Configure Tools

Add MCP tools and integrations

Create Campaign

Launch your first calling campaign

View Analytics

Monitor performance metrics (Coming Soon)

Next Steps

Agent Configuration

Learn how to edit and manage your agents