Mobileye cuts support time 90% with Amazon Bedrock AgentCore

Mobileye, the autonomous driving pioneer with 230M+ deployed chips, faced a growing bottleneck where 66% of support tickets were routine status inquiries requiring engineers to manually navigate 15 clicks across multiple systems. To solve this, they built an AI agent using Amazon Bedrock AgentCore that leverages Anthropic's Claude models through an internal LLM Gateway. The Model Context Protocol (MCP) enabled real-time API access to drive-data platforms, allowing the agent to handle complex inquiries like session status verification, error debugging, and process guidance automatically.

The solution cut response times from 15+ minutes to under 2 minutes while maintaining 95%+ accuracy—eliminating 90% of the manual workload. Mobileye chose Amazon Bedrock AgentCore for its serverless enterprise readiness, built-in observability, and hybrid architecture support that bridges on-premises systems with AWS. The success led to AgentCore becoming a self-service platform for teams company-wide, demonstrating how AI agents can scale support operations without infrastructure overhead or governance trade-offs.