Getting Started with Agents

Learn how to create your first AI agent to automate customer conversations and streamline your sales process.

Learning Resources

📺 Complete Video Guide - New to Agents? Watch our comprehensive step-by-step guide: View Interactive Learning Guide → Complete walkthrough from setup to deployment

📖 Quick Reference - Need a refresher? Follow the text instructions below with photo references for key steps.

Prerequisites

Caution

Not all users have permission to edit pre-created agents. This safeguards optimized prompts and prevents breaking fine-tuned agents.

Before creating your first agent, make sure you have:

• Proper system permissions
• Understanding of prompt engineering principles
• Knowledge of your dealership’s sales process
• Access to the Agents management interface

When to Use Agents

Tip

Agents are perfect for any process with structured conversation flow, clear context, and defined end goals.

Primary Use Cases

Lead Management

• Initial contact and qualification
• Follow-up sequences and nurturing
• Appointment scheduling

Customer Service

• FAQ responses and information requests
• Service appointment booking
• General inquiries and support

Sales Process

• Product recommendations and information
• Objection handling and negotiation
• Closing conversations and next steps

Creating Your First Agent

Any user can create an agent from scratch, and an AI agent can theoretically be built for absolutely any part of the process as long as there is a structured flow to the conversation, context, and an end goal to turn the agent off once it has completed its objective.

Step-by-Step Creation

1. Access the Agents Page

   • Navigate to /agents or use Cmd/Ctrl + K → “Agents”
   • Choose between Grid or List view for better organization

2. Start New Process

   • Click + Create New Process (top right)
   • This opens a blank agent configuration
   • Consider copying from existing similar agents as a starting point

3. Configure the System Prompt

   • Define the agent’s persona and communication style
   • Set up the conversation flow and objectives
   • Add objection handling and FAQ responses
   • Include lead source-specific context

4. Test in Playground

   • Use the interactive testing environment
   • Simulate customer conversations from their perspective
   • Refine responses based on test results

5. Save and Deploy

   • Save your configuration changes
   • Test with real scenarios before full deployment
   • Monitor performance and adjust as needed

How to Create an Agent

On the Agents page, click + Create New Process (top right) to open a blank slate. Proper prompt engineering skills and techniques are recommended before creating your own agent. If an internal team member is creating an agent, start from a pre-existing agent closest to the desired outcome: copy its prompt from the Prompt editor and paste it into the new agent’s Prompt editor, then adapt.

How Agents Work

Technical Foundation

Agents are text-based and built on the AAI Text Model foundation. They integrate seamlessly with your contact management and conversation systems.

Integration Points

Agents can be selected and activated:

• New Contact Creation - Assign appropriate agent based on lead source
• Bulk SMS Campaigns - Apply consistent agent personality across campaigns
• Conversations Tab - Switch agents mid-conversation via process selector dropdown

Response Generation

When a new message arrives:

1. Agent reads entire conversation history
2. Analyzes current context and conversation stage
3. References complete system prompt and knowledge base
4. Generates contextually appropriate response
5. Follows defined conversation flow progression

Complete Agent Workflow

1. Planning Phase

   • Open /agents and choose Grid or List view
   • Identify the agent type needed for your lead source
   • Review similar existing agents for reference

2. Creation/Modification

   • Create new process or edit existing agent
   • Configure system prompt with all required elements
   • Set up objection handling and FAQ responses

3. Testing Phase

   • Use the playground to test conversation flows
   • Try various customer scenarios and responses
   • Refine responses based on testing results

4. Deployment

   • Save your agent configuration
   • Test with small volume of real conversations
   • Monitor performance and customer satisfaction

5. Optimization

   • Analyze conversation outcomes and success rates
   • Update responses based on real-world performance
   • Continuously improve based on customer feedback

6. Publishing (Super Admin Only)

   • Optionally publish successful agents to the Marketplace
   • Share best practices with other dealerships
   • Contribute to the community knowledge base

---

Ready to create your first agent? Start with the Agent Playground to test your ideas, or dive into Agent Configuration to understand the technical details.