Carvana uses Microsoft Azure as the foundation for its AI-driven customer experience. The company combines its AI agent, Sebastian, with an AI-powered quality review platform called CARE (Conversation Analysis Review Engine), both built on Azure AI Foundry and other Azure services. Sebastian guides customers through key steps of the buying and selling journey, handling many standard questions and tasks. CARE then analyzes millions of customer conversations each month to understand what’s working, where customers get stuck, and how both Sebastian and human Customer Advocates can improve. By moving from sampling a small portion of interactions to having 100% visibility into customer conversations, Carvana can: - Identify and fix issues faster - Keep service quality consistent - Continuously refine training for both AI and human teams This approach has led to more efficient, more personalized, and more seamless customer interactions across the entire online car buying and selling process.

Carvana has seen clear, measurable impact from using Sebastian and CARE on Microsoft Azure: - Over the past two years, inbound calls per sale have dropped by more than **45%**. As Sebastian improves, it handles more routine scenarios, so customers need fewer calls to complete their purchase or sale. - This reduction in calls frees Customer Advocates to focus on higher-value work—more complex issues, personalized support, and situations where human judgment matters most. - With **100% visibility** into customer conversations, CARE validates that training and process changes are working. The data shows that advocates are more aligned in how they serve customers, which supports consistent experiences. Overall, the combination of Sebastian and CARE helps Carvana improve customer satisfaction while also increasing operational efficiency and allowing teams to move faster on continuous improvements.

Carvana’s AI customer experience is built on a set of Azure services chosen for scalability, flexibility, and security: - **Azure Kubernetes Service (AKS):** Most of Carvana’s systems run on AKS, which helps the team scale services reliably and manage infrastructure from a single place. This makes it easier to deploy and operate AI-driven applications like CARE. - **Azure AI Foundry and Azure AI Speech:** CARE uses Azure AI Speech within Azure AI Foundry to quickly and accurately transcribe customer conversations. Azure AI Foundry also gives Carvana access to current AI models and workflows, and the company is testing Azure AI Foundry Agent Service to manage multiple agents with minimal code. - **Azure Cosmos DB:** CARE data is stored in Azure Cosmos DB, which provides flexible schemas, automatic scaling of throughput and storage, and high availability. This lets Carvana’s engineering teams move quickly, adjust data models as they learn, and keep performance strong as volumes grow. - **GitHub Copilot:** While not an Azure service, GitHub Copilot is part of the broader Microsoft ecosystem Carvana uses. It helps developers automate repetitive coding tasks so they can focus on more complex engineering work. All of this runs in a secure, compliant Azure environment, which is critical because customer conversations can include sensitive information. Microsoft works with Carvana to ensure that data used by Sebastian and CARE is handled in a compliant and secure way.