What Is My Agent Actually Doing?
Details
### What Is My Agent Actually Doing?
Observing and Tracing AI Agents in Production on AWS
Getting an AI agent to work in a demo is the easy part.
The real challenge begins when that agent reaches production — when it makes an unexpected decision, calls the wrong tool, gets stuck in a workflow, or suddenly consumes far more time and cost than expected.
So, what is your agent actually doing?
In this session, we’ll go beyond simply knowing whether an agent worked and explore how to understand what happened inside a running agent.
We’ll look at how observability can help you trace agent behaviour across reasoning steps, tool calls, model interactions, tokens, latency, and failures — turning otherwise mysterious production issues into something you can investigate and understand.
### What we'll explore
🔹 Tracing AI agents in production
Understand what happens across the different steps of an agentic workflow.
🔹 Understanding how agents fail
Explore three common failure patterns:
- Wrong or unexpected results
- Incomplete or broken workflows
- Runaway latency and cost
🔹 Observability with AWS
See how Amazon Bedrock AgentCore Observability and Amazon CloudWatch can help capture, monitor, and investigate agent behaviour.
🔹 From visibility to trust
Learn why observability isn't just about monitoring systems — it is a critical foundation for AI governance, explainability, reliability, and trust.
### The takeaway
You’ll leave with a practical mental model for designing and operating observable, reliable, and trustworthy AI agents — especially when real users, real workloads, and real consequences enter the picture.
### Who should attend?
This session is designed for anyone building, deploying, or operating AI agents, including:
• AI/ML Engineers
• GenAI Developers
• Software Engineers
• Cloud Architects
• Data Scientists
• DevOps/MLOps Engineers
• AI Leaders
• Technology Enthusiasts
If you are asking, “How do I know what my AI agent is actually doing once it goes into production?” — this session is for you.