Tutorial: Using the AI Assistant

Tutorial: Using the AI Assistant

The Carfluence AI assistant is powered by OpenAI GPT-4o-mini and has full context about your inventory, integrations, and platform configuration.

Getting Started

Open the AI assistant from the sidebar or via the API:

# Create a conversation
POST /api/v1/ai/conversations

# Send a message
POST /api/v1/ai/conversations/:id/messages
{
  "content": "How many active vehicles do I have?"
}

Example Prompts

Inventory Analysis:

• •"How many vehicles do I have in stock?"
• •"What's my average price for used trucks?"
• •"Show me all BMWs that have been on the lot more than 60 days"
• •"Which 5 vehicles have the highest markup?"

Integration Status:

• •"When was my last vAuto sync?"
• •"Are there any integration errors?"
• •"How many records came in from HomeNet today?"

Data Insights:

• •"What's my average days on lot by make?"
• •"Which dealer has the most inventory?"
• •"Compare my new vs used vehicle count this month"

Platform Help:

• •"How do I set up a merge config?"
• •"What API endpoints are available?"
• •"How do I create a transformation rule?"

Tips for Better Results

• 1Be specific: "Show me all used F-150s under $40k" is better than "show me trucks"
• 2Ask follow-ups: The AI remembers conversation context
• 3Request formats: "Show this as a table" or "Give me the API curl command"
• 4Chain questions: Start broad, then drill down into specifics

Conversation Management

# List all conversations
GET /api/v1/ai/conversations

# Continue an existing conversation (maintains context)
POST /api/v1/ai/conversations/:id/messages
{
  "content": "Now filter those to only Texas dealers"
}

Limitations

• •The AI has a daily cost cap ($5/day by default) to prevent runaway charges
• •Complex analytical queries may take a few seconds longer
• •The AI cannot modify data — it's read-only for safety
• •Responses are limited to the data in your schema (multi-tenant isolation applies)