AI agents have become essential in production environments, but their complexity often leaves engineers struggling with observability. Monitoring these agents is critical for debugging failures, identifying performance issues, and managing costs. The tutorial offers a step-by-step guide on deploying an AI agent using Amazon Bedrock AgentCore, combined with observability tools like OpenTelemetry and Grafana Cloud. Amazon Bedrock AgentCore is a managed service designed to streamline the deployment of AI agents. It eliminates the need for server provisioning and integrates with various foundation models. Key features include container-based deployment, built-in security, and support for orchestration frameworks like CrewAI and LangGraph. The tutorial emphasizes using OpenTelemetry for monitoring, which provides insights into LLM API calls, token consumption, and error tracking in agent workflows. OpenLit enhances this process by allowing automatic instrumentation without requiring code changes. The tutorial walks through creating a research assistant agent with CrewAI, detailing the setup process, required configurations, and how to implement observability features. It includes instructions for specifying environment variables to connect with Grafana Cloud for monitoring purposes. By the end of the guide, users will have a fully deployed agent that is instrumented for performance tracking and debugging, enabling better control over AI agent operations.