Mobileye has used Amazon Bedrock AgentCore to build an internal AI support agent for its data collection processing pipeline, according to an AWS case study. The system handles routine ticket status inquiries that previously required engineers to move through multiple backend systems and logs before replying.
The autonomous driving company processes thousands of drive-recording sessions daily, creating frequent questions from engineering and data teams. AWS says the agent cut response times by 90 percent and exceeded 95 percent accuracy targets in the targeted workflow.
A key enabler was the Model Context Protocol, which lets the agent query real-time processing APIs, retrieve logs, and inspect session status during inference. The model access is routed through Mobileye’s internal LLM gateway for governance and quotas, with Bedrock providing foundation models.
The result moved beyond a single support bot. Mobileye turned AgentCore into a self-service platform for other teams to deploy agents. The case shows how enterprises are using agents first on narrow internal workflows where the data, tools, and success criteria are well understood.