Carvana transforms automotive customer experience with AI-powered automation
Carvana, a leading online automotive retailer, sought to deliver an exceptional and seamless car buying and selling experience. By leveraging Microsoft Azure AI Foundry and related services, Carvana implemented both an AI agent, Sebastian, and the proprietary Conversation Analysis Review Engine (CARE). These systems work together to analyze 100% of customer interactions at scale, enabling in-depth quality monitoring and actionable insights. The solution utilized Azure Kubernetes Service (AKS) for scalable infrastructure, GitHub Copilot to accelerate development, Azure AI Speech for transcription, and Azure Cosmos DB for high availability and flexibility in data storage. Tight integration with the Microsoft Azure ecosystem facilitated quick development iterations and simplified management across services. With CARE, Carvana stopped relying on random sampling and achieved full visibility into customer conversations, quickly identifying areas for training and service improvement. Security and compliance were prioritized through Azure’s capabilities, meeting the demands of handling sensitive customer information securely and reliably. The AI-powered review of conversations has improved the effectiveness of both Sebastian (the AI agent) and human Customer Advocates, allowing the latter to focus on complex issues while the AI handles routine ones. This resulted in enhanced efficiency, reduced inbound calls per sale, and more consistent, high-quality customer service. Through this partnership with Microsoft, Carvana is now pioneering the use of multi-agent AI and continually embracing new Azure AI Foundry agent services to further innovate and drive operational excellence in automotive retail.
- Organization
- Carvana
- Industry
- Automotive
- Location
- United States
Reported outcomes
100%
quantified impactQuality & accuracy
Strategic outcomes
Primary read
Use case focus
Showing 3 of 3
- 1AI-powered quality monitoring for customer support in automotive retail
- 2Conversational analytics for customer experience improvement
- 3Automated AI agent for standard support scenarios in automotive sales
- Inability to analyze 100% of customer support conversations, relying on limited sampling.
- High volume of inbound calls per sale, reducing operational efficiency and customer satisfaction.
- Need for seamless integration and rapid development of new AI-powered capabilities.
- Necessity to handle sensitive customer data securely and in compliance with industry standards.
- Developed AI-powered Conversation Analysis Review Engine (CARE) with Azure AI Foundry for full-spectrum conversation analytics.
- Deployed Sebastian, an AI agent, to automate support for standard customer scenarios.
- Utilized Azure Kubernetes Service for scalable service deployments and Cosmos DB for fast, flexible data storage.
- Accelerated feature development with GitHub Copilot and implemented Azure AI Speech for transcription.
- Leveraged the security and compliance features of Microsoft Azure.
- Analyzes 100% of customer interactions, eliminating blind spots in quality monitoring.
- Reduced inbound customer calls per sale by more than 45% over two years.
- Improved customer advocate focus on complex problems, increasing satisfaction and operational efficiency.
- Faster release of AI-powered features through accelerated development cycles with GitHub Copilot and Azure.
Architecture
CARE, Carvana’s AI-powered quality review engine, uses Azure AI Foundry for analytics, Azure Kubernetes Service to ensure scalable deployment, Azure AI Speech for transcription of interactions, Azure Cosmos DB for highly available storage, and GitHub Copilot for accelerated development. Sebastian, the AI agent, integrates with CARE and the backend infrastructure to automate standard customer support scenarios, while insights from CARE continuously improve both the AI agent and human advocates.
Sources & evidence1
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Was this useful?